CN111738144A - Surface water product generation method and system based on Google Earth Engine cloud platform - Google Patents
Surface water product generation method and system based on Google Earth Engine cloud platform Download PDFInfo
- Publication number
- CN111738144A CN111738144A CN202010568232.2A CN202010568232A CN111738144A CN 111738144 A CN111738144 A CN 111738144A CN 202010568232 A CN202010568232 A CN 202010568232A CN 111738144 A CN111738144 A CN 111738144A
- Authority
- CN
- China
- Prior art keywords
- surface water
- water body
- image
- cloud platform
- pixel
- 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
- 239000002352 surface water Substances 0.000 title claims abstract description 129
- 238000000034 method Methods 0.000 title claims abstract description 28
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 62
- 238000012549 training Methods 0.000 claims abstract description 40
- 238000007637 random forest analysis Methods 0.000 claims abstract description 23
- 239000013598 vector Substances 0.000 claims description 14
- 230000001932 seasonal effect Effects 0.000 claims description 11
- 238000012544 monitoring process Methods 0.000 claims description 7
- 238000009360 aquaculture Methods 0.000 claims description 4
- 244000144974 aquaculture Species 0.000 claims description 4
- 238000003384 imaging method Methods 0.000 claims description 4
- 239000013535 sea water Substances 0.000 claims description 4
- 238000005520 cutting process Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000003908 quality control method Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 7
- 238000004519 manufacturing process Methods 0.000 abstract description 4
- 239000003673 groundwater Substances 0.000 abstract 1
- 238000003066 decision tree Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000002310 reflectometry Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000010205 computational analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Astronomy & Astrophysics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a surface water product generation method and system based on a Google Earth Engine cloud platform, wherein the method comprises the steps of acquiring a full-time Landsat8 satellite remote sensing image by utilizing a Google Earth Engine remote sensing big data cloud platform, forming different types of water and non-water typical sample libraries by manually selecting high-precision training samples, constructing characteristic wave bands by adopting multiple indexes such as normalized vegetation indexes, normalized water indexes, terrain gradients and the like and image original wave bands, and constructing a surface water occurrence frequency model by adopting a random forest classifier by combining the high-precision training samples and the characteristic wave bands to realize intelligent extraction of the full-time ground water and realize automatic generation of the surface water product; the water body product generated by the invention has higher space-time resolution, can reflect the dynamic change rule of the surface water body seasons, and can be applied to the automatic production of the surface water body product with long time sequence in national or global scales.
Description
Technical Field
The invention relates to the technical field of remote sensing image intelligent identification, in particular to a surface water product generation method and system based on a Google Earth cloud platform.
Background
Water is the foundation for supporting the survival and development of various lives in the nature, and meanwhile, the water also deeply influences the nature and the human landscape of the earth surface layer. In general, the surface water body refers to the types of the surface of lakes, rivers, reservoirs, cultivation areas and the like in inland areas. Influenced by factors such as natural rainfall, human production and life and the like, the surface water body has obvious seasonal dynamic change characteristics. The satellite remote sensing has the advantages of large scale, low cost, quick revisit and the like, and can quickly and accurately master the space-time dynamic distribution information of the surface water body.
Fast-returning satellites represented by MODIS and SSM/I, AVHRR are applied earlier to develop surface water frequency product mapping, and representative surface water products such as Global Waterpack, GIEMS-D15 and SWAMPS are developed, but the space resolution is low, and the information of the time-space dynamic change of the tiny surface water is difficult to accurately reflect. With Landsat opening and sharing, global earth surface water body drawing steps into the era of high temporal spatial resolution, and series earth surface water body products such as GLCF GIW, GLOWABO, G3WBM, FROM-GLC water mask and the like are developed. The surface water body has obvious seasonal dynamic change characteristics, and the current surface water body products are more than single-time phase products and are difficult to reflect the seasonal dynamic of the water body. Part of surface water products are developed by adopting multi-temporal images, but products developed by utilizing full-time-sequence Landsat remote sensing data are still few.
With the rapid development of artificial intelligence technology and cloud computing, a remote sensing information intelligent service platform represented by Google Earth Engine is created, the platform integrates Landsat series satellite data, MODIS, Sentinel series and other multi-source remote sensing image data, Earth data, atmospheric data and economic data, can realize large-scale image on-line computational analysis, and provides a new opportunity for developing high-space-time-resolution surface water body mapping.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a surface water product generation method based on a Google Earth Engine cloud platform, which can realize the automatic generation of surface water frequency products.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
a surface water product generation method based on a Google Earth Engine cloud platform comprises the following steps:
s1, acquiring full-time Landsat8 remote sensing data and preprocessing the data;
s2, selecting high-precision training sample points from the Landsat8 remote sensing image;
s3, constructing a plurality of image characteristic wave bands and index characteristic wave bands for distinguishing the water body from the non-water body;
s4, constructing a random forest classifier model, and training the random forest classifier model by using the high-precision training sample library selected in the step S2 and the characteristic wave band constructed in the step S3;
s5, carrying out automatic classification and extraction on the surface water body of each scene image according to the random forest classifier model trained in the step S4 to form a full-time surface water body classification data set;
s6, constructing a surface water body appearance frequency model on a pixel scale according to the full-time surface water body classification data set formed in the step S5, and calculating the surface water body appearance frequency of each pixel;
and S7, automatically generating surface water body occurrence frequency products representing the seasonal dynamic rules of the surface water bodies.
Further, the full-time Landsat8 remote sensing data in the step S1 specifically selects a landform 8 surface reflectance product of the full time sequence within a set time, and removes the cloud and the cloud shadow in the image by using a quality control band including different clouds and cloud shadows in the image.
Further, the step S2 is to select a high-precision training sample point from the Landsat remote sensing image, and specifically includes the following sub-steps:
s21, selecting a region with the image coverage area occupying a set proportion of the whole region area;
s22, selecting surface water sample points of various water body types including but not limited to lakes, rivers, reservoirs and pools, near-shore seawater, silt water bodies, aquaculture and the like on each image;
s23, selecting various types of non-water body sample points including but not limited to farmlands, forests, grasslands, cities, bare lands, bushes, ice and snow, clouds, cloud shadows, mountain shadows and wetlands on each image;
s25, acquiring sample points from the image metadata, selecting reference image track number information, and marking image imaging time;
and S26, organizing the training sample points in a spatial point data file format.
Further, the image characteristic bands in step S3 specifically include 6 image characteristic bands from Landsat8 satellite image band 2 to band 7;
the index characteristic wave bands specifically comprise 5 index characteristic wave bands of a normalized vegetation index, a global environment monitoring index, a normalized water body index, a normalized humidity index and a slope.
Further, the training of the random forest classifier model by using the high-precision training sample points selected in step S2 and the feature band constructed in step S3 in step S4 specifically includes:
according to the high-precision training sample points selected in the step S2, pixel values of each characteristic band in the reference image are correspondingly obtained, and a characteristic vector is constructed, which is expressed as:
V′＝V(B2，B3，B4，B5，B6，B7，BNDVI，BGEMI，BNDWI，BNDMI，BSLOPE)
wherein V' represents a feature vector, V (.) represents a feature set, B2，B3，B4，B5，B6，B7Respectively representing the pixel values, B, of the 2 to 7 bands of the satellite image band to the 7 band image band of Landsat8NDVI，BGEMI，BNDWI，BNDMI，BSLOPERespectively representing a normalized vegetation index, a global environment monitoring index, a normalized water body index and a normalized humidity indexAnd pixel values of the slope index characteristic band;
and training and optimizing parameters of the random forest classifier model by using the constructed feature vectors to obtain the optimized random forest classifier model.
Further, the full-time surface water body classification data set formed in the step S5 is expressed as:
S′＝S(A(i，j，t1)，A(i，j，t2)，A(i，j，t3)，...，A(i，j，tn))
wherein S' represents a full-time-sequence surface water body classification result, S (.) represents a full-time-sequence surface water body vector set, and A (i, j, tn) represents tn time phase i row j column surface water body classification results.
Further, the surface water occurrence frequency model in step S6 is expressed as:
FW (i, j) represents the occurrence frequency of the surface water body at the pixel positions of i rows and j columns, t represents the observation order, N represents the total satellite observation number, O (i, j) represents whether the pixel positions of i rows and j columns are observed as the surface water body, if the pixel is observed as the water body, O (i, j) is 1, and if the pixel is observed as the non-water body, O (i, j) is 0; s (i, j) indicates whether or not the pixel in i row and j column is a valid earth observation, and if the pixel is an earth observation, S (i, j) is 1, and if the pixel is an observation of a cloud or a cloud shadow, S (i, j) is 0.
Further, the step S7 is to generate a surface water occurrence frequency product specifically as follows:
in a Google Earth Engine remote sensing big data cloud platform, generating a surface water body appearance frequency product according to the calculated surface water body appearance frequency of each pixel;
carrying out noise removal treatment on the surface water body appearance frequency product by using the gradient of the digital elevation model;
and (5) carrying out inlaying and cutting treatment on the surface water frequency product after the noise is removed.
The invention further provides a surface water product generation system based on the Google Earth cloud platform, based on the surface water product generation method based on the Google Earth cloud platform, and the surface water product generation system generates a surface water appearance frequency product by applying the surface water product generation method.
The invention has the following beneficial effects:
the method comprises the steps of acquiring a full-time Landsat8 satellite remote sensing image by using a Google Earth Engine remote sensing big data cloud platform, forming different types of water bodies and non-water body typical sample libraries by manually selecting high-precision training samples, constructing characteristic wave bands by using multiple indexes such as normalized vegetation indexes, normalized water body indexes and terrain gradients and image original wave bands, combining the high-precision training samples and the characteristic wave bands, realizing full-time Earth surface water body intelligent extraction by using a random forest classifier, constructing an Earth surface water body appearance frequency model, and realizing automatic generation of Earth surface water body products; the water product generated by the invention has higher space-time resolution, can reflect the dynamic change rule of the surface water season, and can be applied to the automatic production of the global scale long-time-sequence surface water product.
Drawings
FIG. 1 is a schematic flow chart of a surface water product generation method based on a Google Earth Engine cloud platform according to the present invention;
FIG. 2 is a schematic diagram of a surface water appearance frequency product generated in an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
In order to better reflect seasonal dynamic change information of surface water bodies, a Google Earth Engine remote sensing big data cloud platform is used for obtaining full-time Landsat8 satellite remote sensing images, high-precision training samples are manually selected to form different types of water bodies and non-water body typical sample libraries, multiple indexes such as normalized vegetation indexes, normalized water body indexes and terrain gradients and image original wave bands are used for constructing characteristic wave bands, a random forest classifier is used for achieving full-time surface water body intelligent extraction by combining the high-precision training samples and the characteristic wave bands, a surface water body occurrence frequency model is constructed, and surface water body product automatic generation is achieved.
As shown in fig. 1, a method for generating a surface water product based on a Google Earth Engine cloud platform provided by the embodiment of the present invention includes the following steps S1 to S7:
s1, acquiring full-time Landsat8 remote sensing data and preprocessing the data;
in the embodiment, the method is based on a Google Earth Engine remote sensing big data cloud platform, and Landsat8 ground surface reflectivity products of a full time sequence within a set time are selected, wherein the set time is usually annual and seasonal, and the geometric positioning accuracy of the products is not lower than one pixel.
According to the invention, the quality control wave band (QA) in the image is adopted, and the wave band comprises mask ranges of different clouds and cloud shadows, so that the clouds and the cloud shadows in the image can be directly removed, and the error separation during the water body extraction is avoided.
S2, selecting high-precision training sample points from the Landsat8 remote sensing image;
in this embodiment, since the accuracy of selecting sample points is directly related to the accuracy of extracting surface water, the method selects a typical area Landsat8 image in a research area, and selects high-accuracy training sample points by means of human-computer interaction and expert experience support, and specifically includes the following steps:
s21, selecting a region with the image coverage area occupying a set proportion of the whole region area;
according to the size of a research area, particularly in a large-scale area, the area of an image coverage area accounts for 5% -10% of the area of the whole area.
S22, selecting a set number of typical sample points on each scene image, wherein the typical sample points are distributed uniformly in space and comprise water body sample points and non-water body sample points of different types;
the method selects 30-50 typical sample points on each scene image, the typical sample points are distributed uniformly in space and comprise different types of water body and non-water body sample points, and the number of the specifically selected sample points is divided according to the complexity of selecting the ground objects of the typical images.
S23, selecting multiple water body type sample points including but not limited to lakes, rivers, reservoirs, near-shore seawater, silt water bodies and aquaculture from each image;
because the spectral differences reflected by the land surface water bodies with different water qualities are large, the selected samples need to cover all the surface water body types of the images aiming at each image, and specifically comprise types of lakes, rivers, pools, near-shore seawater, silt water bodies, aquaculture and the like.
S24, selecting various non-water body sample points including but not limited to farmlands, forests, grasslands, cities, bare lands, bushes, ice and snow, clouds, cloud shadows, mountain shadows and wetlands on each image;
when the non-water body sample is selected, besides the selection of the types of farmlands, forests, cities, bare lands, bushes and the like, the non-water body which is easy to be mixed with the surface water body, such as ice and snow, clouds, cloud shadows, mountain shadows and wetlands, also needs to be selected to a certain extent.
S25, acquiring sample points from the image metadata, selecting reference image track number information, and marking image imaging time;
because the surface water body has seasonal variation, the invention not only needs to record the track number information of the sample point selection reference image for each sample point, but also needs to mark the imaging time of the image, thereby ensuring that the sample point can not influence the training precision of the model because of the error caused by the seasonal fluctuation of the water body. Wherein the sample point selects the reference image track number information to obtain from the image metadata.
And S26, organizing the training sample points in a spatial point data file format.
The training sample points selected by the invention are organized in a Shapefile space point data file format, the file adopts a WGS1984 coordinate system, and the field information of the attribute table is shown in a table 1.
TABLE 1 training sample Point Attribute Table field information
S3, constructing a plurality of image characteristic wave bands and index characteristic wave bands for distinguishing the water body from the non-water body;
in the embodiment, the method realizes effective distinguishing between the water body and the non-water body by constructing a plurality of characteristic wave bands.
Aiming at the characteristics of Landsat8 satellite images, 6 image characteristic wave bands from wave band 2 to wave band 7 of Landsat8 satellite images are selected, but the wave band 1 is not included. The reason is that the waveband is not set in Landsat5 and Landsat7, and the trained model can be suitable for Landsat5 and Landsat7 satellite remote sensing data by selecting waveband 2 to waveband 7, so that the universality of the model is improved.
In addition to the 6 image characteristic bands, 4 index characteristic bands are selected, and specifically include a normalized vegetation index (NDVI), a Global Environment Monitoring Index (GEMI), a normalized water body index (NDWI), a normalized humidity index (NDMI) and a SLOPE (SLOPE).
The calculation formula of the normalized vegetation index is as follows:
where ρ isnirIs Landsat8 wave band 5, rhoredLandsat8 band 4.
The calculation formula of the global environment monitoring index is as follows:
where ρ isnirIs Landsat8 wave band 5, rhoredLandsat8 band 4, adjustment factor.
The calculation formula of the normalized water body index is as follows:
where ρ isgreenIs Landsat8 waveband 3, rhonirLandsat8 band 5.
The normalized humidity index is calculated as:
ρniris Landsat8 wave band 5, rhoswir1Landsat8 band 6.
The calculation formula of the gradient is as follows:
wherein dz/dx is the ratio of elevation z to the x direction, and dz/dy is the ratio of elevation z to the y direction.
In summary, the present invention selects 11 characteristic bands, wherein the image characteristic band is 5, and the index characteristic band is 6.
S4, constructing a random forest classifier model, and performing optimization training on the classifier model by using the high-precision training sample points selected in the step S2 and the characteristic wave bands constructed in the step S3;
in this embodiment, the invention first constructs a random forest classifier model, performs ensemble learning on a plurality of decision trees, and classifies the training data set by constructing a plurality of decision trees, and the model performs bootstrap sampling algorithm on a given training data set D, D { (X)1，Y1)，(X2，Y2)，...(Xm，Ym) Repeatedly iterating for B times by taking X and Y as training feature vectors, and performing pairsBagging at the B th time, wherein B is 1, 2, B, and a sampling set DbTraining a b-th decision tree model Gb (x), selecting an optimal characteristic on a node of the model through multiple times of training to divide left and right subtrees of the decision tree, and regarding classification, using a class which is most votes cast by a plurality of decision tree models as a final class.
According to the high-precision training sample library selected in the step S2, the pixel values of each characteristic wave band in the reference image are correspondingly obtained, and a characteristic vector is constructed and expressed as:
V′＝V(B2，B3，B4，B5，B6，B7，BNDVI，BGEMI，BNDWI，BNDMI，BSLOPE)
wherein V' represents a feature vector, V (.) represents a feature set, B2，B3，B4，B5，B6，B7Respectively representing the pixel values, B, of the 2 to 7 bands of the satellite image band to the 7 band image band of Landsat8NDVI，BGEMI，BNDWI，BNDMI，BSLOPERespectively representing pixel values of characteristic wave bands of a normalized vegetation index, a global environment monitoring index, a normalized water body index, a normalized humidity index and a gradient index;
and training and optimizing parameters of the random forest classifier model by using the constructed feature vectors to obtain the random forest classifier model with better applicability after optimization.
S5, automatically extracting the surface water body of each scene image according to the random forest classifier model optimized in the step S4 to form a full-time surface water body classification data set;
in this embodiment, according to a trained random forest classifier model, the invention performs automatic extraction of surface water for each scene image in the full-time Landsat8 remote sensing data preprocessed in step S1 to form a full-time surface water classification dataset, which is expressed as:
S′＝S(A(i，j，t1)，A(i，j，t2)，A(i，j，t3)，...，A(i，j，tn))
wherein S' represents a full-time-sequence surface water body classification result, S (·) represents a full-time-sequence surface water body vector set, A (i, j, tn) represents a tn-time-phase i row j column surface water body classification result, a water body pixel attribute is marked as 1, and a non-water body pixel attribute is marked as 0.
S6, constructing a surface water body appearance frequency model at a pixel scale according to the full-time surface water body classification data set formed in the step S5, and calculating the surface water body appearance frequency of each pixel;
in this embodiment, based on a full-time-sequence surface water body classification dataset, a land surface water body appearance frequency model is constructed for t-phase i rows and j columns of pixel a (i, j, t), and is expressed as:
FW (i, j) represents the occurrence frequency of the surface water body at the position of the i rows and the j columns of pixels, the permanent water body is 100%, the non-water body is 0%, the seasonal water body is 0-100%, t represents the t-th observation, N represents the total number of satellite observations, O (i, j) represents whether the i rows and the j columns of pixels are observed as the surface water body, if the pixel is observed as the water body, O (i, j) is 1, and if the pixel is observed as the non-water body, O (i, j) is 0; s (i, j) indicates whether or not the pixel in i row and j column is a valid earth observation, and if the pixel is an earth observation, S (i, j) is 1, and if the pixel is an observation of a cloud or a cloud shadow, S (i, j) is 0.
And S7, generating a surface water body occurrence frequency product.
In the embodiment, in the Google Earth Engine remote sensing big data cloud platform, Landsat8 remote sensing surface reflectivity products are selected, a high-precision training sample library is input, a random forest classifier model is constructed and trained, full-time-sequence surface water bodies in a selected area can be rapidly extracted by utilizing the high-performance parallel computing capability of the Google Earth Engine cloud platform, then the surface water body appearance frequency of each pixel is calculated through a surface water body appearance frequency model, and the surface water body appearance frequency is automatically generated.
Then, carrying out noise removal treatment on the products with the frequency of the surface water body by using the gradient of the digital elevation model; the denoising formula is expressed as:
where SLOPE is the SLOPE and FW (i, j) represents the frequency of occurrence of the surface water at i rows and j columns of pixels.
Finally, the surface water frequency product with the noise removed is subjected to inlaying and cutting treatment, and the surface water product in the required range is extracted, as shown in fig. 2.
According to the method, Landsat8 satellite remote sensing images and random forest classifier models in the Google Earth Engine remote sensing big data cloud platform are combined to perform full-time-sequence surface water body intelligent extraction, and meanwhile, a surface water body appearance frequency model is constructed, so that the surface water body product systematic generation can be realized based on the Landsat8 satellite images.
The invention also provides a surface water appearance frequency product generation system based on the surface water appearance frequency product generation method, and the system can automatically generate a surface water appearance frequency product by applying the surface water appearance frequency product generation method.
The system for generating the surface water body product has the characteristic of strong practicability, the surface water body product generated by the system has higher space-time resolution, can accurately reflect the seasonal dynamic change rule of the surface water body, and can be applied to the national or global scale long-time-sequence surface water body product automatic production.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (9)
1. A surface water product generation method and system based on a Google Earth Engine cloud platform are characterized by comprising the following steps:
s1, acquiring a full-time Landsat8 remote sensing image and preprocessing the image;
s2, selecting high-precision training sample points from the Landsat8 remote sensing image;
s3, constructing a plurality of image characteristic wave bands and index characteristic wave bands for distinguishing the water body from the non-water body;
s4, constructing a random forest classifier model, and training and optimizing the random forest classifier model by using the high-precision training sample library selected in the step S2 and the characteristic wave band constructed in the step S3;
s5, automatically extracting the surface water body of the full-time Landsat8 image according to the random forest classifier model optimized in the step S4 to form a full-time surface water body classification data set;
s6, constructing a surface water body appearance frequency model at a pixel scale according to the full-time surface water body classification data set formed in the step S5, and calculating the surface water body appearance frequency of each pixel;
and S7, automatically generating surface water body occurrence frequency products representing the seasonal dynamic rules of the surface water bodies.
2. The Google Earth Engine cloud platform-based surface water product generation method as claimed in claim 1, wherein the Landsat8 surface reflectance product of the full time sequence within the set time is specifically selected from the full time sequence Landsat8 remote sensing data in the step S1, and the cloud shadow in the image are removed by adopting a mask range quality control band including different clouds and cloud shadows in the image.
3. The Google Earth Engine cloud platform-based surface water product generation method of claim 2, wherein the step S2 of selecting a high-precision training sample library from Landsat8 remote sensing images comprises the following sub-steps:
s21, selecting a region with the image coverage area occupying a set proportion of the whole region area;
s22, selecting a set number of typical sample points on each scene image, wherein the typical sample points are uniformly distributed in space and comprise sample points of different types of water bodies and non-water bodies;
s23, selecting multiple water body type sample points including but not limited to lakes, rivers, reservoirs, near-shore seawater, silt water bodies and aquaculture from each image;
s24, selecting various non-water body sample points including but not limited to farmlands, forests, grasslands, cities, bare lands, bushes, ice and snow, clouds, cloud shadows, mountain shadows and wetlands on each image;
s25, acquiring sample points from the image metadata, selecting reference image track number information, and marking image imaging time;
and S26, organizing the training sample points in a spatial point data file format.
4. The method for generating a surface water product based on the Google Earth Engine cloud platform as claimed in claim 3, wherein the image characteristic wave bands in the step S3 specifically include 6 image characteristic wave bands from Landsat8 satellite image wave band 2 to wave band 7;
the index characteristic wave bands specifically comprise 5 index characteristic wave bands of a normalized vegetation index, a global environment monitoring index, a normalized water body index, a normalized humidity index and a slope.
5. The Google Earth Engine cloud platform-based surface water product generation method of claim 4, wherein the training of the random forest classifier model in step S4 is performed by using the high-precision training sample points selected in step S2 and the characteristic wave bands constructed in step S3, and specifically comprises the steps of:
according to the high-precision training sample points selected in the step S2, pixel values of each characteristic band in the reference image are correspondingly obtained, and a characteristic vector is constructed, which is expressed as:
V′＝V(B2，B3，B4，B5，B6，B7，BNDVI，BGEMI，BNDWI，BNDMI，BSLOPE)
whereinV' denotes a feature vector, V (.) denotes a feature set, B2，B3，B4，B5，B6，B7Respectively representing the pixel values, B, of the 2 to 7 bands of the satellite image band to the 7 band image band of Landsat8NDVI，BGEMI，BNDWI，BNDMI，BSLOPERespectively representing pixel values of characteristic wave bands of a normalized vegetation index, a global environment monitoring index, a normalized water body index, a normalized humidity index and a gradient index;
and training and optimizing parameters of the random forest classifier model by using the constructed feature vectors to obtain the optimized random forest classifier model.
6. The Google Earth Engine cloud platform-based surface water product generation method of claim 5, wherein the full-time-sequence surface water vector formed in the step S5 is represented as:
S′＝S(A(i，j，t1)，A(i，j，t2)，A(i，j，t3)，...，A(i，j，tn))
wherein S' represents a full-time-sequence surface water body classification result, S (.) represents a full-time-sequence surface water body vector set, and A (i, j, tn) represents tn time phase i row j column surface water body classification results.
7. The Google Earth Engine cloud platform-based surface water product generation method of claim 6, wherein the surface water appearance frequency model in step S6 is represented as:
FW (i, j) represents the occurrence frequency of the surface water body at the pixel positions of i rows and j columns, t represents the observation order, N represents the total satellite observation number, O (i, j) represents whether the pixel positions of i rows and j columns are observed as the surface water body, if the pixel is observed as the water body, O (i, j) is 1, and if the pixel is observed as the non-water body, O (i, j) is 0; s (i, j) indicates whether or not the pixel in i row and j column is a valid earth observation, and if the pixel is an earth observation, S (i, j) is 1, and if the pixel is an observation of a cloud or a cloud shadow, S (i, j) is 0.
8. The Google Earth Engine cloud platform-based surface water product generation method of claim 7, wherein the step S7 of generating a surface water appearance frequency product specifically comprises:
in a Google Earth Engine remote sensing big data cloud platform, generating a surface water body appearance frequency product according to the calculated surface water body appearance frequency of each pixel;
carrying out noise removal treatment on the surface water body appearance frequency product by using the gradient of the digital elevation model;
and (5) carrying out inlaying and cutting treatment on the surface water frequency product after the noise is removed.
9. A surface water product generation system based on the Google Earth Engine cloud platform, wherein the generation system applies the surface water product generation method according to any one of claims 1 to 8 to generate a surface water appearance frequency product.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010568232.2A CN111738144B (en) | 2020-06-19 | 2020-06-19 | Surface water product generation method and system based on Google Earth Engine cloud platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010568232.2A CN111738144B (en) | 2020-06-19 | 2020-06-19 | Surface water product generation method and system based on Google Earth Engine cloud platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111738144A true CN111738144A (en) | 2020-10-02 |
| CN111738144B CN111738144B (en) | 2021-08-31 |
Family
ID=72651866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010568232.2A Active CN111738144B (en) | 2020-06-19 | 2020-06-19 | Surface water product generation method and system based on Google Earth Engine cloud platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111738144B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112733947A (en) * | 2021-01-14 | 2021-04-30 | 中国水利水电科学研究院 | Tailing dam space distribution identification system and method |
| CN112800973A (en) * | 2021-01-29 | 2021-05-14 | 宁波大学 | Spartina alterniflora extraction method based on vegetation phenological feature decision |
| CN112966657A (en) * | 2021-03-25 | 2021-06-15 | 中国科学院空天信息创新研究院 | Remote sensing automatic classification method for large-scale water body coverage |
| CN112990657A (en) * | 2021-02-06 | 2021-06-18 | 首都师范大学 | Method for evaluating wetland degradation through long-time satellite remote sensing |
| CN112991351A (en) * | 2021-02-23 | 2021-06-18 | 新华三大数据技术有限公司 | Remote sensing image semantic segmentation method and device and storage medium |
| CN113408467A (en) * | 2021-07-01 | 2021-09-17 | 中国科学院东北地理与农业生态研究所 | Coastal culture pond intelligent extraction method based on Sentinel-2 satellite images and cloud platform |
| CN113420780A (en) * | 2021-05-24 | 2021-09-21 | 中国科学院地理科学与资源研究所 | Culture pond extraction method based on remote sensing spatial-temporal spectral feature fusion |
| CN113610882A (en) * | 2021-04-23 | 2021-11-05 | 华北水利水电大学 | Surface water body drawing method and device, electronic equipment and storage medium |
| CN114120137A (en) * | 2021-10-19 | 2022-03-01 | 桂林理工大学 | Wetland element space-time evolution monitoring method based on time sequence main remote sensing image |
| CN114882359A (en) * | 2022-05-07 | 2022-08-09 | 中国科学院空天信息创新研究院 | Soybean planting area extraction method and system based on vegetation index time series spectrum characteristics |
| CN115965878A (en) * | 2022-12-07 | 2023-04-14 | 四川大学 | Inversion monitoring method for suspended sediment of Tibet plateau river based on GEE |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102565778A (en) * | 2011-12-12 | 2012-07-11 | 中国科学院遥感应用研究所 | Relative radiometric correction method for automatically extracting pseudo-invariant features for remote sensing image |
| CN106125091A (en) * | 2016-06-22 | 2016-11-16 | 广州地理研究所 | The city impervious surface Remotely sensed acquisition method that line spectrum solution is mixed |
| CN107730496A (en) * | 2017-10-26 | 2018-02-23 | 武汉大学 | A kind of multi-temporal remote sensing image building change detecting method based on image blocks |
| CN108388832A (en) * | 2018-01-11 | 2018-08-10 | 福州大学 | A kind of conceding the land to forestry automatic identifying method based on multiple timings index variation tendency |
| CN108647738A (en) * | 2018-05-17 | 2018-10-12 | 中国科学院遥感与数字地球研究所 | Global Scale remote sensing image water body intelligent extract method based on multi index option |
| CN109543552A (en) * | 2018-10-29 | 2019-03-29 | 华南农业大学 | Paddy growth region drafting method and system based on growth period multi-feature extraction |
| CN110033499A (en) * | 2019-04-21 | 2019-07-19 | 南京林业大学 | A kind of regional water body drafting method based on Landsat OLI remote sensing image |
| CN110276797A (en) * | 2019-07-01 | 2019-09-24 | 河海大学 | A kind of area of lake extracting method |
| CN110685747A (en) * | 2019-08-27 | 2020-01-14 | 中国矿业大学(北京) | Remote sensing extraction method for coal mining subsidence water body of high diving space |
| CN110738187A (en) * | 2019-10-24 | 2020-01-31 | 中国科学院城市环境研究所 | quick remote sensing estimation method for dynamic change of lake area based on Google Earth Engine |
| CN110751727A (en) * | 2019-10-30 | 2020-02-04 | 南京林业大学 | Synthetic image construction method based on Landsat long-time sequence |
| CN110852262A (en) * | 2019-11-11 | 2020-02-28 | 南京大学 | Agricultural land extraction method based on time sequence top-grade first remote sensing image |
| CN111046613A (en) * | 2019-02-21 | 2020-04-21 | 北京大学 | Optimal river channel calculation method based on path tracking and river network extraction method based on multi-temporal remote sensing image |
| CN111259876A (en) * | 2020-05-06 | 2020-06-09 | 中国水利水电科学研究院 | Radar data water body information extraction method and system based on land surface water body product |
-
2020
- 2020-06-19 CN CN202010568232.2A patent/CN111738144B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102565778A (en) * | 2011-12-12 | 2012-07-11 | 中国科学院遥感应用研究所 | Relative radiometric correction method for automatically extracting pseudo-invariant features for remote sensing image |
| CN106125091A (en) * | 2016-06-22 | 2016-11-16 | 广州地理研究所 | The city impervious surface Remotely sensed acquisition method that line spectrum solution is mixed |
| CN107730496A (en) * | 2017-10-26 | 2018-02-23 | 武汉大学 | A kind of multi-temporal remote sensing image building change detecting method based on image blocks |
| CN108388832A (en) * | 2018-01-11 | 2018-08-10 | 福州大学 | A kind of conceding the land to forestry automatic identifying method based on multiple timings index variation tendency |
| CN108647738A (en) * | 2018-05-17 | 2018-10-12 | 中国科学院遥感与数字地球研究所 | Global Scale remote sensing image water body intelligent extract method based on multi index option |
| CN109543552A (en) * | 2018-10-29 | 2019-03-29 | 华南农业大学 | Paddy growth region drafting method and system based on growth period multi-feature extraction |
| CN111046613A (en) * | 2019-02-21 | 2020-04-21 | 北京大学 | Optimal river channel calculation method based on path tracking and river network extraction method based on multi-temporal remote sensing image |
| CN110033499A (en) * | 2019-04-21 | 2019-07-19 | 南京林业大学 | A kind of regional water body drafting method based on Landsat OLI remote sensing image |
| CN110276797A (en) * | 2019-07-01 | 2019-09-24 | 河海大学 | A kind of area of lake extracting method |
| CN110685747A (en) * | 2019-08-27 | 2020-01-14 | 中国矿业大学(北京) | Remote sensing extraction method for coal mining subsidence water body of high diving space |
| CN110738187A (en) * | 2019-10-24 | 2020-01-31 | 中国科学院城市环境研究所 | quick remote sensing estimation method for dynamic change of lake area based on Google Earth Engine |
| CN110751727A (en) * | 2019-10-30 | 2020-02-04 | 南京林业大学 | Synthetic image construction method based on Landsat long-time sequence |
| CN110852262A (en) * | 2019-11-11 | 2020-02-28 | 南京大学 | Agricultural land extraction method based on time sequence top-grade first remote sensing image |
| CN111259876A (en) * | 2020-05-06 | 2020-06-09 | 中国水利水电科学研究院 | Radar data water body information extraction method and system based on land surface water body product |
Non-Patent Citations (4)
| Title |
|---|
| TROY STERNBERG等著: "《Arid Land Systems: Sciences and Societies》", 19 August 2019 * |
| 冯美晨: "《黄图高原水旱第地冬小麦长势动态监测及产量品质遥感估测》", 31 October 2015, 中国农业大学出版社 * |
| 朱春娇: "基于markov和clue-s模型的浦东新区土地利用变化趋势分析", 《中国优秀硕士学位论文全文数据库经济与管理科学辑》 * |
| 熊春宝等: "《地理信息系统原理与工程应用》", 31 August 2014, 天津大学出版社 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112733947B (en) * | 2021-01-14 | 2021-11-09 | 中国水利水电科学研究院 | Tailing dam space distribution identification system and method |
| CN112733947A (en) * | 2021-01-14 | 2021-04-30 | 中国水利水电科学研究院 | Tailing dam space distribution identification system and method |
| CN112800973A (en) * | 2021-01-29 | 2021-05-14 | 宁波大学 | Spartina alterniflora extraction method based on vegetation phenological feature decision |
| CN112800973B (en) * | 2021-01-29 | 2021-08-27 | 宁波大学 | Spartina alterniflora extraction method based on vegetation phenological feature decision |
| CN112990657A (en) * | 2021-02-06 | 2021-06-18 | 首都师范大学 | Method for evaluating wetland degradation through long-time satellite remote sensing |
| CN112990657B (en) * | 2021-02-06 | 2021-12-07 | 首都师范大学 | Method for evaluating wetland degradation through long-time satellite remote sensing |
| CN112991351A (en) * | 2021-02-23 | 2021-06-18 | 新华三大数据技术有限公司 | Remote sensing image semantic segmentation method and device and storage medium |
| CN112991351B (en) * | 2021-02-23 | 2022-05-27 | 新华三大数据技术有限公司 | Remote sensing image semantic segmentation method and device and storage medium |
| CN112966657A (en) * | 2021-03-25 | 2021-06-15 | 中国科学院空天信息创新研究院 | Remote sensing automatic classification method for large-scale water body coverage |
| CN113610882A (en) * | 2021-04-23 | 2021-11-05 | 华北水利水电大学 | Surface water body drawing method and device, electronic equipment and storage medium |
| CN113420780A (en) * | 2021-05-24 | 2021-09-21 | 中国科学院地理科学与资源研究所 | Culture pond extraction method based on remote sensing spatial-temporal spectral feature fusion |
| CN113408467B (en) * | 2021-07-01 | 2022-04-29 | 中国科学院东北地理与农业生态研究所 | Coastal culture pond intelligent extraction method based on Sentinel-2 satellite images and cloud platform |
| CN113408467A (en) * | 2021-07-01 | 2021-09-17 | 中国科学院东北地理与农业生态研究所 | Coastal culture pond intelligent extraction method based on Sentinel-2 satellite images and cloud platform |
| CN114120137A (en) * | 2021-10-19 | 2022-03-01 | 桂林理工大学 | Wetland element space-time evolution monitoring method based on time sequence main remote sensing image |
| CN114882359A (en) * | 2022-05-07 | 2022-08-09 | 中国科学院空天信息创新研究院 | Soybean planting area extraction method and system based on vegetation index time series spectrum characteristics |
| CN115965878A (en) * | 2022-12-07 | 2023-04-14 | 四川大学 | Inversion monitoring method for suspended sediment of Tibet plateau river based on GEE |
| CN115965878B (en) * | 2022-12-07 | 2023-08-29 | 四川大学 | GEE-based inversion monitoring method for suspended sediment in Qinghai-Tibet plateau river |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111738144B (en) | 2021-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111738144B (en) | Surface water product generation method and system based on Google Earth Engine cloud platform | |
| Mao et al. | National wetland mapping in China: A new product resulting from object-based and hierarchical classification of Landsat 8 OLI images | |
| Kattenborn et al. | UAV data as alternative to field sampling to map woody invasive species based on combined Sentinel-1 and Sentinel-2 data | |
| Sylvain et al. | Mapping dead forest cover using a deep convolutional neural network and digital aerial photography | |
| Toosi et al. | Comparing different classification algorithms for monitoring mangrove cover changes in southern Iran | |
| Novelli et al. | Performance evaluation of object based greenhouse detection from Sentinel-2 MSI and Landsat 8 OLI data: A case study from Almería (Spain) | |
| Muñoz et al. | From local to regional compound flood mapping with deep learning and data fusion techniques | |
| Klemas | Remote sensing of coastal wetland biomass: An overview | |
| KR100678395B1 (en) | System and method for real time position correction of geometric information data using satellite and aerospace image | |
| Lassalle et al. | Deep learning-based individual tree crown delineation in mangrove forests using very-high-resolution satellite imagery | |
| CN112712535A (en) | Mask-RCNN landslide segmentation method based on simulation difficult sample | |
| Zhu et al. | Estimating and mapping mangrove biomass dynamic change using WorldView-2 images and digital surface models | |
| Choi et al. | UAV-based land cover mapping technique for monitoring coastal sand dunes | |
| Luo et al. | VHR GeoEye-1 imagery reveals an ancient water landscape at the Longcheng site, northern Chaohu Lake Basin (China) | |
| Wagner | Geographic information systems and glacial environments | |
| Broussard III et al. | Quantifying vegetation and landscape metrics with hyperspatial unmanned aircraft system imagery in a coastal oligohaline marsh | |
| Liu et al. | Thick cloud removal under land cover changes using multisource satellite imagery and a spatiotemporal attention network | |
| Wenger et al. | Multisenge: a multimodal and multitemporal benchmark dataset for land use/land cover remote sensing applications | |
| van Geffen et al. | SiDroForest: A comprehensive forest inventory of Siberian boreal forest investigations including drone-based point clouds, individually labelled trees, synthetically generated tree crowns and Sentinel-2 labelled image patches | |
| Kupková et al. | Towards reliable monitoring of grass species in nature conservation: Evaluation of the potential of UAV and PlanetScope multi-temporal data in the Central European tundra | |
| Ghosh et al. | Change detection of tropical mangrove ecosystem with subpixel classification of time series hyperspectral imagery | |
| Cresson et al. | Orfeo ToolBox Applications | |
| Olthof et al. | Medium resolution land cover mapping of Canada from SPOT 4/5 data | |
| Jawak et al. | Validation of high-density airborne LiDAR-based feature extraction using very high resolution optical remote sensing data | |
| Fayad et al. | Vision Transformers, a new approach for high-resolution and large-scale mapping of canopy heights |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 20 Chegongzhuang West Road, Haidian District, Beijing 100038 Applicant after: China Institute of Water Resources and Hydropower Research Address before: 100038, Fuxing Road 1, Beijing, Haidian District Applicant before: China Institute of Water Resources and Hydropower Research |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |