Characteristic spectrum-based Sargassum horneri gold tide and Enteromorpha green tide remote sensing classification and identification method

Abstract

The invention belongs to the field of ocean remote sensing information, and particularly relates to a characteristic spectrum-based Sargassum horneri gold tide and Enteromorpha green tide remote sensing classification and identification method. Comprising the following steps: 1) acquiring a sentinel No.2 remote sensing image of a set sea area, and preprocessing the sentinel No.2 remote sensing image; 2) calculating a normalized difference vegetation index (NDVI) of the preprocessed remote sensing image, and extracting a floating macroalgae image according to a calculation result; 3) performing non-water interference factor elimination on the floating macroalgae image by using a triple mask method to obtain a macroalgae bloom area; and 4) based on the macroalgae bloom area, carrying out classification and identification on the Sargassum horneri gold tide and the Enteromorpha green tide. The invention provides an improved spectrum ratio gradient SRG-1 algorithm so as to enhance the distinguishing capability of two algae spectrums, and aims to establish a large algae bloom classification extraction method considering both precision and efficiency so as to provide technical support for monitoring and prevention and control of algae bloom disasters.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of marine remote sensing information, and particularly relates to a method for remote sensing classification and identification of chrysochromulina and enteromorpha based on characteristic spectrum. BACKGROUND

[0002] The existing single algal bloom identification method cannot meet the precise monitoring requirement. Since enteromorpha and chrysochromulina have different spectral characteristics (for example, chrysochromulina has a characteristic absorption valley at 645 nm, and the reflection peak of enteromorpha at 705 nm is more significant), it is expected to realize the automatic differentiation of the two types of algal blooms by fusing the index model of spectral characteristics, and to provide technical support for the monitoring and early warning of large algal blooms.

[0003] For the differentiation of enteromorpha green tide and chrysochromulina gold tide, the existing method based on the reflectance slope index (SRG, such as formula 1) of green and red light bands can realize the classification and identification of the two types of large algal blooms.

[0004]

[0005] Wherein, R rc_green and R rc_red respectively represent the Rayleigh-corrected reflectance of the green light band (560 nm) and the red light band (665 nm). In order to reduce the influence of seawater background, the Rrc value used in the formula needs to be subtracted from the Rrc value of the adjacent seawater. When the SRG value is positive, it indicates that the target is chrysochromulina; when the SRG value is negative, it indicates that the target is enteromorpha.

[0006] However, the SRG method still has some limitations in practical application. First, the spatial heterogeneity of seawater pixel reflectance is low, and the global mean calculation of the background value in the SRG index will weaken the regional difference characteristics; second, although dynamic calculation of local seawater background value can improve the accuracy, it will increase the calculation complexity, and the background value in the formula is easily offset in the subtraction operation. More importantly, the general Sentinel-2 Level-2A (L2A) atmospheric correction product uses the Sen2Cor atmospheric correction algorithm, which is based on the atmospheric radiation transfer model developed by LibRadtran laboratory, and is mainly designed for land surface reflectance assumption (such as dark vegetation or known reflectance target), without considering the atmospheric correction requirements of water area, and has systematic defects in the application of water environment. SUMMARY

[0007] The present application is based on the measured Enteromorpha and Chrysochromulina spectrometric data, combined with the Sentinel-2 MSI band characteristics, combined with the L1C apparent reflectance data, and an improved spectral ratio gradient SRG-1 (spectral ratio gradient index-1) algorithm is proposed to enhance the spectral differentiation ability of the two algae, aiming to establish a large algae algal bloom classification extraction method considering accuracy and efficiency, and to provide technical support for algal bloom disaster monitoring and prevention.

[0008] The technical scheme adopted by the present application to achieve the above-mentioned purposes is:

[0009] A Chrysochromulina and Enteromorpha remote sensing classification and identification method based on characteristic spectrum, comprising the following steps:

[0010] 1) Obtain a Sentinel-2 remote sensing image of a set sea area, and pre-process it;

[0011] 2) Calculate the normalized vegetation index NDVI of the pre-processed remote sensing image, and extract the floating macroalgae image according to the calculation result;

[0012] 3) Use a triple mask method to eliminate non-water body interference factors from the floating macroalgae image to obtain a macroalgae algal bloom area;

[0013] 4) Classify and identify Chrysochromulina and Enteromorpha based on the macroalgae algal bloom area.

[0014] The step 1) pre-processes the remote sensing image, specifically:

[0015] Use the public cloud removal dataset mask to remove cloud and cirrus pixels on the image.

[0016] The step 2) is specifically:

[0017] Calculate the NDVI of the remote sensing image, and the image composed of pixel points with NDVI greater than or equal to 0.2 is taken as the floating macroalgae, wherein the NDVI is specifically:

[0018]

[0019] Wherein, R NIR is the near-infrared band reflectivity, corresponding to the 7 or 8 infrared band of Sentinel-2, R RED is the red band reflectivity, corresponding to the 4th band of Sentinel-2.

[0020] The step 3) is specifically:

[0021] Based on the annual Sentinel-2 image, calculate the normalized difference water index NDWI, and use NDWI greater than or equal to 0.2 as a threshold for dynamic water area mask processing;

[0022] Remove land boundaries based on known coastline vector data;

[0023] Based on the global seabed topography database, seabed topography masking is performed using water depth thresholds.

[0024] The normalized difference water index NDWI is specifically:

[0025] NDWI=(R GREEN -R NIR ) / (R GREEN +R NIR )

[0026] Among them, R GREEN is the green band reflectivity, corresponding to Sentinel-2 selecting the 3rd band, R NIR For near-infrared reflectivity, select band 8 for Sentinel-2.

[0027] The step 4) is specifically as follows:

[0028] Based on the large algae bloom areas, the minimum SRG-1 index of each time item of each month was calculated. The algae bloom areas with SRG-1 ≥ -0.28 were regarded as the copper algae gold tide, and the rest were regarded as the enteromorpha green tide.

[0029] The SRG-1 index is specifically:

[0030]

[0031] Among them, R RED is the red band atmospheric top reflectance, corresponding to the 4th band of Sentinel-2, R GREEN is the reflectance of the top atmosphere in the green band, corresponding to the third band of Sentinel-2, λ RED is the central wavelength of the red band, λ GREEN is the center wavelength of the green band.

[0032] A remote sensing classification and identification system for the golden tide of copper algae and the green tide of Enteromorpha based on characteristic spectra, including:

[0033] Image preprocessing module, used to obtain Sentinel-2 remote sensing images of a specified sea area and perform preprocessing on them;

[0034] The floating macroalgae image extraction module is used to calculate the normalized vegetation index NDVI of the pre-processed remote sensing image and extract the floating macroalgae image based on the calculation result;

[0035] A triple mask processing module is used to eliminate non-water interference factors from floating macroalgae images using a triple mask method to obtain macroalgae bloom areas;

[0036] The classification and identification module is used for classifying and identifying copper algae gold tide and green tide of Enteromorpha based on large-scale algal bloom area.

[0037] A device for remotely classifying and identifying copper algae gold tide and green tide of Enteromorpha based on characteristic spectrum, comprising a memory and a processor; the memory is used for storing a computer program; the processor is used for realizing the method for remotely classifying and identifying copper algae gold tide and green tide of Enteromorpha based on characteristic spectrum when the computer program is executed.

[0038] A computer readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the method for remotely classifying and identifying copper algae gold tide and green tide of Enteromorpha based on characteristic spectrum is realized.

[0039] The present application has the following advantages and benefits:

[0040] 1. High classification and identification accuracy of green tide and gold tide: the present application generates detection points randomly in different regions of the South Yellow Sea at six different phases, and uses a confusion matrix to quantitatively determine the identification accuracy of the method and the index for green tide, gold tide and non-algal bloom area. Based on the method, the classification and identification performance of the two types of macroalgae algal blooms is generally good, and has high consistency (overall accuracy: 83%, Kappa coefficient: 68%). There is a certain missed detection phenomenon in the identification of green tide (recall rate: 51%), which is mainly due to the exclusion of part of the low-density green tide characteristic pixels by the threshold value of NDVI>0.2; however, the identification accuracy remains at a high level (precision: 79%), indicating that the misjudgment rate is low. In contrast, the recall rate (>80%) and the precision rate (>80%) of the gold tide identification are both excellent, indicating that the method has better comprehensive performance for gold tide extraction.

[0041] 2. High-precision classification and identification ability of green tide and gold tide: the SRG-1 index (Spectral Ratio Gradient Index-1) proposed in the present application significantly improves the classification accuracy of copper algae gold tide and green tide of Enteromorpha by optimizing the reflectance gradient calculation of the red light band (Sentinel 2 No. 4 band, 665 nm) and the green light band (Sentinel 2 No. 3 band, 560 nm). Based on the confusion matrix analysis of 902 verification points, the overall accuracy of the method for classifying the two types of algal blooms reaches 83%, and the Kappa coefficient is 68%, which has obvious advantages over traditional methods. It is particularly worth noting that the identification performance of the copper algae gold tide is particularly outstanding, with a recall rate and a precision rate both exceeding 80%, fully embodying the superiority of the algorithm in the identification of specific algal blooms.

[0042] 3. The SRG-1 algorithm optimization significantly improves the calculation efficiency: The existing SRG index relies on the Rayleigh-corrected reflectance, but the atmospheric correction model relied on by the general atmospheric correction reflectance product (e.g., Sentinel-2 L2A data) is not optimized for water areas, resulting in systematic errors. The present application directly uses the atmospheric surface reflectance for gradient calculation, not only improving the stability of the algorithm, but also significantly simplifying the data processing process. This improvement makes the algorithm no longer need complex local seawater background value calculation, greatly reduces the calculation complexity while ensuring accuracy, and realizes significant efficiency improvement.

[0043] 4. High-efficiency business processing capability: The present application effectively removes background noise through triple masking, combined with normalized difference water index (NDWI≥0.2), coastline vector data and seabed topography water depth threshold (depth<-6m), effectively eliminates cloud, land, shallow sea vegetation and ships, etc. Interference, avoid the interference of coastal vegetation around the islands; The entire process has been realized automatic processing based on large-scale remote sensing cloud computing platform, which can complete the preprocessing and classification of all data in the Yellow Sea and East China Sea within 1 hour, meeting the demand of large-scale near real-time monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 Method flowchart of the present application;

[0045] Figure 2 Comparison chart of Enteromorpha and Chrysochromulina spectrum measurement curves;

[0046] Figure 3 Comparison chart of Enteromorpha and Chrysochromulina spectrum measurement curves;

[0047] Figure 4 Extraction result chart of Enteromorpha green tide and Chrysochromulina gold tide in the Yellow Sea and East China Sea from May 26 to 30, 2020. DETAILED DESCRIPTION

[0048] The present application will be further described in detail below in combination with the drawings and examples.

[0049] As Figure 1 shown, a Chrysochromulina gold tide and Enteromorpha green tide remote sensing classification and identification method based on characteristic spectrum includes the following steps:

[0050] (1) Extraction of floating macroalgae. Obtain the Sentinel-2 multispectral imager Level-1C (L1C) remote sensing image of the Yellow Sea and East China Sea, and combine the cloud removal dataset (Cloud Score+S2_HARMONIZED V1) provided by the Google Earth Engine platform to remove cloud and cirrus pixels on the image.

[0051] (2) Extracting floating macroalgae by calculating NDVI and setting the threshold range of NDVI ≥ 0.2.

[0052] To calculate the spatial distribution dataset of macroalgae blooms in long time series, the NDVI index was calculated (as shown in formula 2), the near-infrared (NIR) band used Sentinel-2 band 7 (782.5 nm), and the red (RED) band used Sentinel-2 band 4 (664.5 nm) to jointly construct the NDVI index.

[0053] The normalized vegetation index (NDVI) was used to extract macroalgae blooms, and the calculation formula is shown in formula (2).

[0054]

[0055] NDVI is related to the photosynthetic capacity and energy absorption capacity of the vegetation canopy, and this algorithm has been widely used in the study of vegetation coverage and land use on land. Many scholars have used NDVI to extract harmful algae blooms and have achieved good results.

[0056] (3) Elimination of non-water body interference factors. A triple masking method was used, first, the normalized difference water index (NDWI) was calculated based on annual Sentinel-2 images, the calculation formula is shown in formula (3), and the NDWI was calculated using band 3 (560 nm) and band 8A (864 nm) data, with NDWI greater than or equal to 0.2 as the threshold for dynamic water area masking; second, high-precision coastline vector data was integrated to remove the land boundary; third, based on the GEBCO global seafloor topography database, the seafloor topography masking was processed by setting a depth threshold (depth <-6m) to remove the interference of coastal vegetation near islands.

[0057] NDWI = (R B3 -R B8A ) / (R B3 +R B8A )(3)

[0058] (4) Classification and identification of Chrysochromulina gold tide and Enteromorpha green tide. Based on the macroalgae bloom area (pixel set) extracted in the previous step, the SRG-1 minimum value of each month was calculated, and Chrysochromulina gold tide was determined when SRG-1 ≥-0.28, and Enteromorpha green tide was determined when SRG-1 <-0.28.

[0059] Based on the SRG index proposed by Min et al. (2019), we developed the SRG-1 index to distinguish green and golden blooms (as shown in Equation 4). This method is based on the Sentinel-2 MSL1C dataset without Rayleigh correction, and the calculation method is as follows:

[0060]

[0061] In the above formula, R665 and R560 are the TOA reflectance of Sentinel-2 at 665 nm and 560 nm bands, respectively. B4 and R B3 are the TOA reflectance of Sentinel-2 at 665 nm and 560 nm bands, respectively.

[0062] Finally, through multiple masks, we obtained the spatial distribution data of C. prolifera golden blooms, excluding noise information such as land, islands, artificial shorelines, and ships.

[0063] Figure 2 For comparison of Enteromorpha and C. prolifera spectral measurement curves (orange: C. prolifera, green: Enteromorpha), we conducted in-situ spectral measurements on C. prolifera and Enteromorpha with different fresh weights, and obtained the spectral graph as shown in Figure 2 The horizontal axis represents the band information, and the vertical axis represents the spectral reflectance. By comparing the spectral characteristics of the two algae blooms, we can see that the average reflectance of Enteromorpha at the 560 nm green band is 0.016, which is higher than the average reflectance of C. prolifera (0.012); while at the 660 nm red band, the average reflectance of Enteromorpha (0.007) is significantly lower than that of C. prolifera (0.011). The reverse difference in red and green band reflectance leads to a significant difference in the slope of the spectral curve of the two algae, providing a basis for distinguishing Enteromorpha and C. prolifera based on multispectral remote sensing data.

[0064] Figure 3 Comparison of green and golden bloom pixel SRG-1 and NDVI characteristics in the Yellow Sea and East China Sea in 2021 and 2022. To evaluate the adaptability of the SRG-1 algorithm and its threshold, we selected green and golden blooms in the Yellow Sea and East China Sea in 2021 and 2022 as the verification object, and constructed four verification scenarios based on Sentinel-2 L2A data (Golden bloom in the South Yellow Sea on May 1, 2021, green tide in the South Yellow Sea on June 1, 2021, golden bloom in the East China Sea on April 16, 2022, and green tide in the South Yellow Sea on June 21, 2022). After threshold segmentation (0.2 as the critical value), the target pixels were classified and identified, and the frequency distribution characteristics of different categories of SRG-1 index were counted, as shown in Fig. 3. It can be seen that in the high-density algal bloom area with NDVI≥0.2, with -0.28 as the discrimination threshold of SRG-1 index, Enteromorpha green tide (light green) and C. prolifera golden tide (red) can be effectively distinguished; but in the low value interval of NDVI<0.2, the SRG-1 index of the two types of algae blooms shows significant overlapping distribution, indicating that the algorithm has limited ability to identify low-density algae spots.

[0065] Figure 4 The extraction results of Enteromorpha and Sargassum blooms in the Yellow Sea and East China Sea from May 26 to 30, 2020 were based on the construction method of the present application. The satellite remote sensing data obtained in some sea areas of the Yellow Sea and East China Sea from May 26 to 30, 2020 were analyzed. Since the revisit period of Sentinel-2 satellite is 5 days, the complete coverage of the sea area of the Yellow Sea and East China Sea was realized by integrating the image data of four adjacent orbits (the collection dates were from May 26 to 30, respectively). Figure 4 In the global true color composite image shown in a, the left South Yellow Sea experimental area includes two kinds of algal blooms, Enteromorpha green tide (green) and Sargassum gold tide (red), while the East China Sea experimental area shows only Sargassum gold tide (red). For the typical mixed algal bloom area in the South Yellow Sea, local enlargement ( Figure 4 b) was carried out, and in the true color image of this sea area, two types of floating algae showing green and red colors existed at the same time. Through further distinction by SRG-1 algorithm, Enteromorpha green tide ( Figure 4 c green area) and Sargassum gold tide ( Figure 4 c yellow area) could be successfully distinguished, the separation boundary of the two types of algal blooms was clear, and the spatial distribution characteristics were highly consistent with the visual interpretation results.

Claims

1. A remote sensing classification and identification method for the golden tide of copper algae and the green tide of Enteromorpha based on characteristic spectrum, characterized in that: The following steps are involved: 1) Obtain Sentinel-2 remote sensing images of the designated sea area and preprocess them; 2) Calculate the normalized difference vegetation index (NDVI) of the preprocessed remote sensing image and extract the floating macroalgae image based on the calculation result; 3) Using the triple mask method to eliminate non-water interference factors from the floating macroalgae image, the macroalgae bloom area was obtained; 4) Classify and identify the golden tide of copper algae and the green tide of Enteromorpha based on the large algae bloom areas.

2. The remote sensing classification and identification method of the copper algae gold tide and enteromorpha green tide based on characteristic spectrum according to claim 1 is characterized in that: In the step 1), the remote sensing image is preprocessed, specifically: Use the publicly available cloud removal dataset masks to remove cloud and cirrus pixels from the image.

3. The remote sensing classification and identification method of the copper algae gold tide and enteromorpha green tide based on characteristic spectrum according to claim 1 is characterized in that: The step 2) is specifically as follows: Calculate the NDVI of the remote sensing image, and use the image composed of pixels with NDVI ≥ 0.2 as floating macroalgae, where the NDVI is specifically: Among them, R NIR is the near-infrared reflectivity, corresponding to Sentinel-2, which selects infrared bands such as 7 or 8, R RED is the red band reflectivity, and the corresponding Sentinel-2 selects the 4th band.

4. The remote sensing classification and identification method of the copper algae gold tide and enteromorpha green tide based on characteristic spectrum according to claim 1, characterized in that: The step 3) is specifically as follows: The Normalized Difference Water Index (NDWI) was calculated based on the annual Sentinel-2 images, and dynamic water masking was performed using an NDWI greater than or equal to 0.2 as the threshold. Remove land boundaries based on known coastline vector data; Based on the global seabed topography database, seabed topography masking is performed using water depth thresholds.

5. The remote sensing classification and identification method of the copper algae gold tide and the enteromorpha green tide based on characteristic spectrum according to claim 4 is characterized in that: The normalized difference water index NDWI is specifically: NDWI(R GREEN -R NIR ) / (R GREEN +R NIR ) Among them, R GREEN is the green band reflectivity, corresponding to Sentinel-2 selecting the 3rd band, R NIR For near-infrared reflectivity, select band 8 for Sentinel-2.

6. The remote sensing classification and identification method of the copper algae gold tide and enteromorpha green tide based on characteristic spectrum according to claim 1, characterized in that: The step 4) is specifically as follows: Based on the large algae bloom areas, the minimum SRG-1 index of each time item of each month was calculated. The algae bloom areas with SRG-1 ≥ -0.28 were regarded as the copper algae gold tide, and the rest were regarded as the enteromorpha green tide.

7. The remote sensing classification and identification method of the copper algae gold tide and enteromorpha green tide based on characteristic spectrum according to claim 6, characterized in that: The SRG-1 index is specifically: Among them, R RED is the red band atmospheric top reflectance, corresponding to the 4th band of Sentinel-2, R GREEN is the reflectance of the top atmosphere in the green band, corresponding to the third band of Sentinel-2, λ RED is the central wavelength of the red band, λ GREEN is the center wavelength of the green band.

8. A remote sensing classification and identification system for the golden tide of copper algae and the green tide of Enteromorpha based on characteristic spectrum, characterized in that: include: Image preprocessing module, used to obtain Sentinel-2 remote sensing images of a specified sea area and perform preprocessing on them; The floating macroalgae image extraction module is used to calculate the normalized vegetation index NDVI of the pre-processed remote sensing image and extract the floating macroalgae image based on the calculation result; A triple mask processing module is used to eliminate non-water interference factors from floating macroalgae images using a triple mask method to obtain macroalgae bloom areas; The classification and identification module is used to classify and identify the golden tide of copper algae and the green tide of Enteromorpha based on the large algae bloom area.

9. A remote sensing classification and identification device for the golden tide of copper algae and the green tide of Enteromorpha based on characteristic spectrum, characterized in that: It comprises a memory and a processor; the memory is used to store a computer program; the processor is used to implement a remote sensing classification and identification method for the gold tide of copper algae and the green tide of Enteromorpha based on characteristic spectrum as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by the processor, the remote sensing classification and identification method of the copper algae gold tide and the enteromorpha green tide based on characteristic spectrum as described in any one of claims 1 to 7 is implemented.

Images