CN114821348A - Sea ice drawing method - Google Patents
Sea ice drawing method Download PDFInfo
- Publication number
- CN114821348A CN114821348A CN202110121503.4A CN202110121503A CN114821348A CN 114821348 A CN114821348 A CN 114821348A CN 202110121503 A CN202110121503 A CN 202110121503A CN 114821348 A CN114821348 A CN 114821348A
- Authority
- CN
- China
- Prior art keywords
- ice
- cloud
- pixel
- classification
- water
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/20—Drawing from basic elements, e.g. lines or circles
- G06T11/206—Drawing of charts or graphs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Remote Sensing (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a sea ice mapping method, belongs to the technical field of north sea ice mapping and monitoring, and solves the problem of low sea ice detection precision of an existing ice-water mixing area and an ice or water area covered by clouds. The method is applied to the north pole and comprises the following steps: acquiring and preprocessing image data of a plurality of MODISL1B acquired every day in a research area; classifying each preprocessed image data to obtain a classification result matrix of each image data; the categories of the classification include sea ice, water, and cloud; each pixel in the classification result matrix stores a class label corresponding to the pixel; acquiring the total number of non-cloud types and the mode of the non-cloud types of pixels corresponding to the same position in a plurality of classification result matrixes, and obtaining a daily water and ice distribution map based on the total number of the non-cloud types and the mode of the non-cloud types; and fusing the distribution maps of the daily water and ice to obtain a daily synthetic ice map.
Description
Technical Field
The invention relates to the technical field of arctic sea ice mapping and monitoring, in particular to a sea ice mapping method.
Background
Sea ice is an important component of the freezing ring, and has profound influence on the research of global climate change, biodiversity and the like. Particularly, with global climate change, polar climate warming, and the opening of an arctic channel, etc., the demand for sea ice information is rapidly increasing.
The traditional sea ice product mostly depends on the data of a passive microwave radiometer, is not influenced by cloud layers, can form daily global coverage observation, but has low spatial resolution which is mostly data of dozens of kilometers, and is difficult to use for fine analysis and navigation. The SAR data image has the characteristics of all weather and high spatial resolution, can provide high-quality sea ice information, but has longer revisit period and smaller coverage range, and is difficult to provide a large-scale and high-time-resolution sea ice product. Although the optical satellite sensor is influenced by cloud cover and illumination, the optical satellite sensor can still acquire abundant sea ice information due to high time resolution and spatial resolution, for example, the MODIS sensor of TERRA and AQUA satellites in the United states can perform multiple observations in polar regions every day, and the spatial resolution of the acquired data is far higher than that of passive microwave data. By comprehensively analyzing the multi-temporal data, all or part of the sea ice information of the polar region can be acquired. The MOD29 monoscopic sea ice product provided by the American national space agency based on MODIS data can help a user to acquire the sea ice condition of the polar region to a certain extent, but the product often has a misjudgment phenomenon when the ice detection is carried out on ice and water mixed regions and ice or water regions with cloud coverage. This is mainly because the cloud mask data used by the sea ice product provided by NASA is the MOD35 product, which is a cloud product for the world, and the recognition accuracy is poor in polar regions, especially in ice regions covered by clouds, thin clouds, or fog. The product tends to underestimate the cloud cover on sea ice and overestimate the cloud cover on open waters.
Therefore, how to improve the detection accuracy of ice and water mixing areas and ice or water areas covered by clouds is an urgent problem to be solved.
Disclosure of Invention
In view of the foregoing analysis, an embodiment of the present invention is directed to providing a sea ice mapping method, so as to solve the problem of low sea ice detection accuracy in the existing ice and water mixing area and ice or water area with cloud coverage.
The invention provides a sea ice mapping method, which is applied to the north pole and comprises the following steps:
acquiring and preprocessing image data of a plurality of MODIS L1B acquired daily in a research area;
classifying each preprocessed image data to obtain a classification result matrix of each image data; the categories of the classification include sea ice, water, and cloud; each pixel in the classification result matrix stores a class label corresponding to the pixel;
acquiring the total number of non-cloud types and the mode of the non-cloud types of pixels corresponding to the same position in a plurality of classification result matrixes, and obtaining a daily water and ice distribution map based on the total number of the non-cloud types and the mode of the non-cloud types;
and fusing the distribution maps of the daily water and ice to obtain a daily synthetic ice map.
On the basis of the scheme, the invention also makes the following improvements:
further, the daily water profile is obtained by performing the following operations:
for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a first threshold, the value of the pixel is the mode of the non-cloud classes;
and extracting the pixel with the water as the class label, and obtaining a daily water distribution map when the values of other pixels are null.
Further, the daily ice profile is obtained by performing the following operations:
for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a second threshold, the value of the pixel is the mode of the non-cloud classes;
extracting pixels with the category labels of sea ice from the ice distribution map, wherein the values of other pixels are null, and obtaining the distribution map of the ice per day;
wherein the second threshold is greater than the first threshold.
Further, the fusing the daily water and ice distribution maps to obtain a daily synthetic ice map comprises:
respectively comparing the values of the pixels at each corresponding position in the distribution maps of the water and the ice every day, and if the values are consistent, keeping the values unchanged; if the pixel values are inconsistent, the value of the pixel is the non-empty mode of the set neighborhood of the current pixel in the distribution map of water and ice every day; and the values of the other pixels are class labels corresponding to the clouds, so that a daily synthetic ice map is formed.
Further, the method further comprises: a weekly synthetic ice map was obtained based on seven consecutive days of sea ice maps:
acquiring the total number and the mode of non-cloud categories of pixels corresponding to the same position in the ice picture for seven consecutive days;
if the total number of the non-cloud categories of the current pixel is not 0, the value of the current pixel is the non-cloud category mode of the current pixel, and if the number of the sea ice and the water in the non-cloud categories is equal, the value of the current pixel is the category label corresponding to the water;
if the total number of the non-cloud classes of the current pixel is 0, the value is taken as a class label corresponding to the cloud;
thereby forming a weekly synthetic ice pattern.
Further, a classification result matrix of each image data is obtained by performing the following operations:
and extracting a characteristic wave band corresponding to each preprocessed image data, verifying that the classification model after passing the verification is classified based on the extracted characteristic wave band, and obtaining a classification result matrix of each image data.
Further, the classification model after the verification is passed is obtained by performing the following operations:
obtaining MODIS L1B historical image data of a plurality of time stages in a research area and carrying out batch preprocessing;
extracting a plurality of sample data corresponding to each classification type from the historical image data after batch preprocessing, and acquiring a type label corresponding to each sample data;
constructing a prior pixel sample library based on the sample data and the class label corresponding to the sample data;
selecting a training sample and a verification sample from the prior pixel sample library, training a classification model based on the characteristic wave band of each sample data in the training sample and the class label corresponding to the characteristic wave band, verifying the classification accuracy of the trained classification model based on the characteristic wave band of each sample data in the verification sample and the class label corresponding to the characteristic wave band, and if the classification accuracy exceeds a classification accuracy threshold, passing the verification so as to obtain the classification model passing the verification.
Further, the characteristic band includes: data corresponding to MODIS reflectivity bands 7, 2 and 1; NDSI, and MRELBP.
Further, the classification categories corresponding to the clouds further comprise sub-categories of blue cloud, white cloud and red cloud; each sub-category corresponds to a corresponding sub-category label.
Further, the preprocessing comprises geometric correction, radiometric calibration, sun zenith angle correction and land masking.
Compared with the prior art, the invention can realize at least one of the following beneficial effects:
the invention provides a sea ice drawing method, which has the following beneficial effects:
firstly, aiming at the problem that ice and cloud are easy to be confused all the time on the basis of realizing automatic classification of images in a research area, the invention combines the classification results of multiple images acquired every day by utilizing the arctic region cloud motion characteristic and a ground feature multi-temporal judgment method, synthesizes a daily ice map on the basis, and thus effectively solves the problem that ice and cloud are easy to be confused;
second, the present invention also forms a weekly synthetic ice pattern based on the daily synthetic ice pattern;
thirdly, aiming at the problems of low spatial resolution and the like of the traditional sea ice product, the invention constructs a prior pixel sample library for polar region object classification based on the constructed arctic multi-temporal remote sensing image, and automatically realizes the classification of the images in the research area by utilizing a random forest algorithm of multi-feature band fusion. Compared with the traditional MODIS sea ice product, the spatial resolution and accuracy of the classification result are greatly improved. Meanwhile, the invention also improves the selected characteristic wave band and the classification category in the classification process, so that the classification precision is effectively improved.
Additional features, bands and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
Drawings
The drawings are only for purposes of illustrating particular embodiments and are not to be construed as limiting the invention, wherein like reference numerals are used to designate like parts throughout.
FIG. 1 is a flow chart of a method for making sea ice according to embodiment 1 of the present invention;
FIG. 2 is a schematic diagram of the synthetic process of the daily synthetic ice map provided in example 1 of the present invention;
fig. 3(a) and (B) are both false color images of B721 band combination provided in embodiment 2 of the present invention, wherein the acquisition times of fig. 3(a) and (B) are 7/3 days in 2018 and 8/3 days in 2019, respectively;
fig. 3(c) and (d) are the classification results of the images of fig. 3(a) and (b) respectively according to the MFLFRF algorithm in embodiment 2 of the present invention;
fig. 3(e) and (f) are respectively MOD29 sea ice products corresponding to fig. 3(a) and (b) in embodiment 2 of the present invention.
Fig. 4(a) is a daily synthetic ice map obtained based on Terra star for 8 months and 1 day 2019 provided by example 2 of the present invention;
FIG. 4(b) weekly synthetic ice maps obtained on Terra and Aqua stars on days 8, 1-7 in 2019.
Detailed Description
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate preferred embodiments of the invention and together with the description, serve to explain the principles of the invention and not to limit the scope of the invention.
Example 1
The invention discloses a sea ice mapping method, a flow chart is shown in figure 1, the method is applied to the north pole, and the method comprises the following steps:
step S1: acquiring and preprocessing image data of a plurality of MODIS L1B (wherein the L1B data are generated by decompressing, positioning and calibrating MODIS original data and are level 1 product data of MODIS) acquired daily in a research area;
preferably, the preprocessing comprises geometric correction, radiometric calibration, sun zenith angle correction, land mask correction and the like; it is emphasized that, by performing the land mask processing, the preprocessed image data only contains ice, water and cloud; therefore, the categories classified in step S2 include sea ice, water, and cloud.
In this embodiment, the land mask processing process is: extracting a land mask of a research area based on MODIS land cover type annual global 500 m products (MCD12Q1 products); in addition, the processes of geometric correction, radiometric calibration and solar zenith angle correction can be realized by adopting the existing mode, and are not described again here.
Step S2: classifying each preprocessed image data to obtain a classification result matrix of each image data; the categories of the classification include sea ice, water, and cloud; each pixel in the classification result matrix stores a class label corresponding to the pixel;
preferably, the specific process of the step is as follows: extracting a characteristic wave band corresponding to each preprocessed image data, verifying that the passed classification model is classified based on the extracted characteristic wave band, and obtaining a classification result matrix of each image data;
in consideration of the difference between the spectral and texture characteristic bands of sea ice and cloud, the characteristic band selected in this embodiment includes: data corresponding to the MODIS reflectivity bands 7, 2 and 1 (namely, short wave infrared band, near infrared band and red band of the MODIS), wherein the data of the three bands can be directly obtained based on image data; NDSI (Normalized Difference Snow and Ice Index), and MRELBP (Median Robust Extended Local Binary Pattern), NDSI and MRELBP data need to be obtained by processing image data, and the processing process adopts the existing mode, and is not described herein again.
Before step S2 is executed, the classification model after verification is required to be obtained in advance, and the specific process is as follows:
(1) obtaining MODIS L1B historical image data of a plurality of time stages in a research area and carrying out batch preprocessing; preferably, in arctic sea areas, the surface features are of a single type, mainly including sea ice, water and clouds; the feature of the ground feature is relatively fixed, and is relatively less affected by factors such as seasons, and the method is suitable for identifying the ground feature through modes such as sample migration, and therefore, in the embodiment, multi-view images (over 100 views) at different time stages such as a freezing period and an ice-thawing period are selected for batch preprocessing, and sample data is selected from the preprocessed images.
(2) Extracting a plurality of sample data corresponding to each classification type from the historical image data after batch preprocessing, and acquiring a type label corresponding to each sample data;
(3) constructing a prior pixel sample base based on the sample data and the corresponding category label;
(4) selecting a training sample and a verification sample from the prior pixel sample library, training a classification model based on the characteristic wave band of each sample data in the training sample and the class label corresponding to the characteristic wave band, verifying the classification accuracy of the trained classification model based on the characteristic wave band of each sample data in the verification sample and the class label corresponding to the characteristic wave band, and if the classification accuracy exceeds a classification accuracy threshold, passing the verification so as to obtain the classification model passing the verification.
Consider that the color of the clouds in the study area appear as blue, white, red, etc.; therefore, in order to better realize classification, the classification category corresponding to the cloud is further divided into three sub-categories, namely blue cloud, white cloud and red cloud, according to the color characteristics of the cloud; each subcategory corresponds to a corresponding subcategory label. Accordingly, the selected training samples and verification samples should also include the sample data corresponding to each sub-category.
In the process of selecting the training samples and the verification samples, the training samples and the verification samples are selected from multi-scene images acquired in different time of a research area according to the principle of uniform distribution of the samples. Illustratively, 80% of samples in the constructed a priori pixel sample library are used as training samples, the remaining 20% of samples are used as verification samples, and model training is performed by using a random forest (MFLFRF) algorithm. The number of decision trees ntree in the random forest is set to be 400-500, and the number of variables in each level is set to be 4-6.
Through the above process, a classification result matrix of each image data can be obtained.
Consider that there may be misclassifications in the classification process, for example, misclassification of a small fraction of blue clouds and cloud shadows as sea ice. However, the cloud and cloud shadow covering the sea ice are moving, and in the same region of the arctic, a plurality of MODIS images can be acquired in one day (generally, at least 6 observations are performed, 9 observations are performed in most cases, and at least 4 observations are performed in the special region of the arctic edge). Therefore, the influence caused by misclassification can be reduced by further processing the classification result matrix corresponding to a plurality of pieces of image data observed in the same day. The specific procedure is as described in steps S3 and S4.
Step S3: acquiring the total number of non-cloud categories (namely the number of times of distinguishing water or sea ice in the classification, which is represented by N) and the mode of the non-cloud categories of the pixels corresponding to the same position in the classification result matrixes, and obtaining a daily water and ice distribution map based on the total number of the non-cloud categories and the mode of the non-cloud categories;
through a large number of experiments, the classification precision of water is the highest in the classification process of the embodiment. Thus, the following steps are formed:
step S31: for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a first threshold, the value of the pixel is the mode of the non-cloud classes; if the total number of the non-cloud classes is 0, the value of the pixel is a class label corresponding to the cloud; thereby obtaining a daily composite classification map M1; and only extracting the pixel with the class label of water in the daily composite classification map M1, wherein other pixels are null, and obtaining the daily water distribution map.
Step S32: for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a second threshold, the value of the pixel is the mode of the non-cloud classes; if the total number of the non-cloud classes is 0, the value of the pixel is a class label corresponding to the cloud; obtaining a daily synthetic classification chart M2, extracting only the pixel of which the class label is ice in the daily synthetic classification chart M2, and obtaining a daily ice distribution chart if other pixels are empty;
the second threshold is greater than the first threshold, because a few cloud areas are easily mistakenly classified into ice in the classification result, the mistakenly classified cloud areas are filtered out in a mode of increasing the threshold, and the second threshold is set to be 3 according to experience; since the accuracy of the recognition of water is high in relation to the recognition of sea ice, the threshold value may be somewhat lower, and the first threshold value is set to 1 empirically.
It should be noted that, considering that the classification accuracy of water is the highest, in the process of obtaining the daily composite classification maps M1 and M2, if the numbers of sea ice and water in the non-cloud classes are equal, the value of the current pixel is the class label corresponding to water.
Step S4: and fusing the distribution maps of the daily water and ice to obtain a daily synthetic sea ice map.
Preferably, the values of the pixels at each corresponding position in the distribution maps of the water and the ice every day are respectively compared, and if the values are consistent, the values are kept unchanged; if the number of the ice and the water in the non-empty categories of the set neighborhood is equal, the value of the current pixel is the category label corresponding to the water. (i.e. when the number of the ice and the water in the neighborhood is equal, the judgment result is the label corresponding to the water, because the classification precision of the water is higher than that of the sea ice); and the values of the rest pixels are class labels corresponding to the clouds, so that a daily synthetic ice map is formed. A schematic of the daily synthetic ice map synthesis process is shown in FIG. 2.
On the basis of obtaining the daily synthetic ice map, the embodiment may further include the following steps:
step S5: a weekly synthetic ice map was obtained based on a seven-day continuous sea ice mapping:
acquiring the total number and the mode of non-cloud categories of pixels corresponding to the same position in the ice picture for seven consecutive days; if the total number of the non-cloud categories of the current pixel is not 0, the value of the current pixel is the non-cloud category mode of the current pixel, and if the number values of ice and water in the non-cloud categories are consistent, the value of the current pixel is a category label corresponding to the water; if the total number of the non-cloud classes of the current pixel is 0, the value is taken as a class label corresponding to the cloud; thereby forming a weekly synthetic ice pattern.
In summary, compared with the prior art, the sea ice mapping method provided by the embodiment has the following beneficial effects:
firstly, aiming at the problem that ice and cloud are easy to be confused all the time on the basis of realizing automatic classification of images in a research area, the invention combines the classification results of multiple images acquired every day by utilizing the arctic region cloud motion characteristic and a ground feature multi-temporal judgment method, synthesizes a daily ice map on the basis, and thus effectively solves the problem that ice and cloud are easy to be confused;
secondly, the invention also forms a weekly synthetic ice map based on the daily synthetic ice map;
thirdly, aiming at the problems of low spatial resolution and the like of the traditional sea ice product, the invention constructs a prior pixel sample library for polar region object classification based on the constructed arctic multi-temporal remote sensing image, and automatically realizes the classification of the images in the research area by utilizing a random forest algorithm of multi-feature band fusion. Compared with the traditional MODIS sea ice product, the spatial resolution and accuracy of the classification result are greatly improved. Meanwhile, the invention also improves the selected characteristic wave band and the classification category in the classification process, so that the classification precision is effectively improved.
Example 2
The specific embodiment 2 of the present invention is used to verify the accuracy of the sea ice mapping method in the embodiment 1 of the present invention; in particular, the amount of the solvent to be used,
in this embodiment, a classification model training is performed on a constructed prior sample library for a preprocessed original image based on a multi-feature band fused random forest (MFLFRF) algorithm, and a single-scene image is automatically classified in batches by using a classification model passing verification, and the image classification result is displayed by taking two single-scene images as an example, as shown in fig. 3; wherein, fig. 3(a) and (B) are false color images combined by B721 wave band, and the collection time is 7/3 days in 2018 and 8/3 days in 2019 respectively; FIGS. 3(c) and (d) are the classification results of the images of FIGS. 3(a) and (b) by using the MFLFRF classification algorithm; fig. 3(e) and (f) are respectively MOD29 sea ice products corresponding to fig. 3(a) and (b).
In the B721 band combination mode, the identification of blue cloud is shown (as the position marked by the box in fig. 3 (a)), the region in the MOD29 ice map is mistakenly identified as sea ice, and the classification model in the above embodiment can be used to accurately identify blue cloud in the region, but there is still a small part of cloud shadow region being mistakenly identified. For the identification of ice water gaps (as indicated by the box in fig. 3 (b)), the MOD29 ice map product cannot accurately identify water regions in sea ice, and the classification model in the above embodiment (e.g., based on the MFLFRF algorithm) can accurately identify sea ice and water regions. In conclusion, the MFLFRF classification algorithm has higher identification precision in the identification process of sea ice, water and cloud.
The daily and weekly synthetic ice plots obtained based on the procedure in example 1 are shown in figure 4; wherein, fig. 4(a) is a daily synthetic ice map obtained based on Terra star on 8/1/2019; FIG. 4(b) is a weekly synthetic ice map based on Terra and Aqua stars, obtained on 8 months 1-7 days 2019. As can be seen from fig. 4(a) and (b), the arctic region is affected by the cloud layer, most of the daily synthetic ice map is covered by the cloud layer, and only a small part of the daily synthetic ice map is a clear sky region. Most areas in the synthesized ice chart in each week are clear sky areas, but some areas are covered by cloud layers for 7 days continuously, and the cloud layer areas are shown.
Those skilled in the art will appreciate that all or part of the flow of the method implementing the above embodiments may be implemented by a computer program, which is stored in a computer readable storage medium, to instruct related hardware. The computer readable storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (10)
1. A method for mapping sea ice, characterized in that it is applied to the arctic, comprising the following steps:
acquiring and preprocessing image data of a plurality of MODIS L1B acquired daily in a research area;
classifying each preprocessed image data to obtain a classification result matrix of each image data; the categories of the classification include sea ice, water, and cloud; each pixel in the classification result matrix stores a class label corresponding to the pixel;
acquiring the total number of non-cloud types and the mode of the non-cloud types of pixels corresponding to the same position in a plurality of classification result matrixes, and obtaining a daily water and ice distribution map based on the total number of the non-cloud types and the mode of the non-cloud types;
and fusing the distribution maps of the daily water and ice to obtain a daily synthetic ice map.
2. Sea ice mapping method according to claim 1, wherein the daily water profile is obtained by performing the following operations:
for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a first threshold, the value of the pixel is the mode of the non-cloud classes;
and extracting the pixel with the water as the class label, and obtaining a daily water distribution map when the values of other pixels are null.
3. A method for sea ice mapping according to claim 2, wherein said daily ice profile is obtained by performing the following operations:
for each pixel in the classification result matrix, if the total number of the non-cloud classes is greater than a second threshold, the value of the pixel is the mode of the non-cloud classes;
extracting pixels with category labels of sea ice from the ice distribution map, wherein the values of other pixels are null, and obtaining the daily ice distribution map;
wherein the second threshold is greater than the first threshold.
4. The method for mapping sea ice according to any one of claims 1-3, wherein said fusing said daily water and ice profiles to obtain a daily synthetic ice map comprises:
respectively comparing the values of the pixels at each corresponding position in the distribution maps of the water and the ice every day, and if the values are consistent, keeping the values unchanged; if the pixel values are inconsistent, the value of the pixel is the non-empty mode of the set neighborhood of the current pixel in the distribution map of water and ice every day; and the values of the other pixels are class labels corresponding to the clouds, so that a daily synthetic ice map is formed.
5. The method for sea ice mapping according to claim 4, further comprising: a weekly synthetic ice map was obtained based on seven consecutive days of sea ice maps:
acquiring the total number of non-cloud categories and the mode of the non-cloud categories of pixels corresponding to the same position in the sea ice chart for seven consecutive days;
if the total number of the non-cloud categories of the current pixel is not 0, the value of the current pixel is the non-cloud category mode of the current pixel, and if the number of the sea ice and the water in the non-cloud categories is equal, the value of the current pixel is the category label corresponding to the water;
if the total number of the non-cloud classes of the current pixel is 0, the value is taken as a class label corresponding to the cloud;
thereby forming a weekly synthetic ice pattern.
6. The sea ice mapping method according to claim 1, wherein the classification result matrix of each image data is obtained by performing the following operations:
and extracting a characteristic wave band corresponding to each preprocessed image data, verifying that the classification model after passing the verification is classified based on the extracted characteristic wave band, and obtaining a classification result matrix of each image data.
7. The sea ice mapping method according to claim 6, wherein the verified classification model is obtained by performing the following operations:
obtaining MODIS L1B historical image data of a plurality of time stages in a research area and carrying out batch preprocessing;
extracting a plurality of sample data corresponding to each classification type from the historical image data after batch preprocessing, and acquiring a type label corresponding to each sample data;
constructing a prior pixel sample base based on the sample data and the corresponding category label;
selecting a training sample and a verification sample from the prior pixel sample library, training a classification model based on the characteristic wave band of each sample data in the training sample and the class label corresponding to the characteristic wave band, verifying the classification accuracy of the trained classification model based on the characteristic wave band of each sample data in the verification sample and the class label corresponding to the characteristic wave band, and if the classification accuracy exceeds a classification accuracy threshold, passing the verification so as to obtain the classification model passing the verification.
8. A method for sea ice mapping according to claim 6 or 7,
the characteristic band includes: data corresponding to MODIS reflectivity bands 7, 2 and 1; NDSI, and MRELBP.
9. The sea ice mapping method of claim 1 or 7, wherein the classification categories corresponding to the clouds further include sub-categories of blue cloud, white cloud and red cloud; each sub-category corresponds to a corresponding sub-category label.
10. The method for sea ice mapping according to claim 1, wherein said preprocessing comprises geometric correction, radiometric calibration, sun zenith angle correction, and land masking.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121503.4A CN114821348A (en) | 2021-01-28 | 2021-01-28 | Sea ice drawing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110121503.4A CN114821348A (en) | 2021-01-28 | 2021-01-28 | Sea ice drawing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114821348A true CN114821348A (en) | 2022-07-29 |
Family
ID=82525597
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110121503.4A Pending CN114821348A (en) | 2021-01-28 | 2021-01-28 | Sea ice drawing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114821348A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115830446A (en) * | 2022-11-25 | 2023-03-21 | 中国水利水电科学研究院 | Dynamic water product fusion method, device, equipment and readable storage medium |
| CN116030362A (en) * | 2023-02-15 | 2023-04-28 | 山东省科学院海洋仪器仪表研究所 | High-precision sea ice concentration inversion method |
| CN116452985A (en) * | 2023-02-21 | 2023-07-18 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
-
2021
- 2021-01-28 CN CN202110121503.4A patent/CN114821348A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115830446A (en) * | 2022-11-25 | 2023-03-21 | 中国水利水电科学研究院 | Dynamic water product fusion method, device, equipment and readable storage medium |
| CN116030362A (en) * | 2023-02-15 | 2023-04-28 | 山东省科学院海洋仪器仪表研究所 | High-precision sea ice concentration inversion method |
| CN116452985A (en) * | 2023-02-21 | 2023-07-18 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
| CN116452985B (en) * | 2023-02-21 | 2023-10-31 | 清华大学 | Surface water monitoring method, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wagner et al. | Individual tree crown delineation in a highly diverse tropical forest using very high resolution satellite images | |
| US8594375B1 (en) | Advanced cloud cover assessment | |
| CN114821348A (en) | Sea ice drawing method | |
| Li et al. | High-spatial-resolution surface and cloud-type classification from MODIS multispectral band measurements | |
| Huo et al. | Supervised spatial classification of multispectral LiDAR data in urban areas | |
| KR100678395B1 (en) | System and method for real time position correction of geometric information data using satellite and aerospace image | |
| KR102540762B1 (en) | Reservoir monitoring method using satellite informations | |
| Ursani et al. | Fusion of textural and spectral information for tree crop and other agricultural cover mapping with very-high resolution satellite images | |
| Kuzmin et al. | Automatic segment-level tree species recognition using high resolution aerial winter imagery | |
| Jain et al. | Expansion of a glacial lake, Tsho Chubda, Chamkhar Chu Basin, Hindukush Himalaya, Bhutan | |
| Liu et al. | An object-oriented daytime land-fog-detection approach based on the mean-shift and full lambda-schedule algorithms using EOS/MODIS data | |
| Lee et al. | Pre-trained feature aggregated deep learning-based monitoring of overshooting tops using multi-spectral channels of GeoKompsat-2A advanced meteorological imagery | |
| Katoh et al. | Application of high-resolution airborne data using individual tree crowns in Japanese conifer plantations | |
| Wakabayashi et al. | Flooded rice paddy detection using sentinel-1 and planetscope data: a case study of the 2018 spring flood in West Java, Indonesia | |
| Wang et al. | Multi-source evidence data fusion approach to detect daily distribution and coverage of Ulva prolifera in the Yellow Sea, China | |
| Yadav et al. | Urban tree canopy detection using object-based image analysis for very high resolution satellite images: A literature review | |
| Iannone et al. | Proba-V cloud detection Round Robin: Validation results and recommendations | |
| Kakooei et al. | VHR semantic labeling by random forest classification and fusion of spectral and spatial features on Google Earth Engine | |
| Shi et al. | Comparing pixel-based random forest and the object-based support vector machine approaches to map the quasi-circular vegetation patches using individual seasonal fused GF-1 imagery | |
| CN112036246B (en) | Construction method of remote sensing image classification model, remote sensing image classification method and system | |
| Jayasekare et al. | Hybrid method for building extraction in vegetation-rich urban areas from very high-resolution satellite imagery | |
| Lefebvre et al. | Vineyard identification and characterization based on texture analysis in the helderberg basin (South Africa) | |
| Jun et al. | Computer identification of multispectral satellite cloud imagery | |
| Park et al. | Unconstrained approach for isolating individual trees using high-resolution aerial imagery | |
| Kanjir et al. | Application of object based approach to heterogeneous land cover/use |
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 |