Method and device for determining optimal vegetation index of spartina alterniflora

Abstract

The invention provides a method and a device for determining an optimal vegetation index of spartina alterniflora. The method provided by the invention comprises the following steps: acquiring hyperspectral image data of spartina alterniflora; determining a target wave band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data in combination with the spectral characteristics of various vegetation indexes; for each observation point, respectively calculating each vegetation index under the observation point based on the reflectivity information under the target wave band; acquiring field sampling data of the spartina alterniflora; the field sampling data comprises vegetation coverage data of spartina alterniflora under a plurality of field sampling points; for each observation point, matching the vegetation index under the observation point with the vegetation coverage data under the corresponding field sampling point, and establishing a matching relationship; and for each vegetation index, calculating the correlation between the vegetation index under each observation point and the vegetation coverage under the corresponding field sampling point based on the matching relationship, and screening the optimal vegetation index based on the correlation.

Description

Technical Field

[0001] This application relates to the technical field of remote sensing information processing, and particularly relates to a method and device for determining the optimal vegetation index of Spartina alterniflora. Background Art

[0002] Island ecosystems play an important role in maintaining global plant community types and species diversity, but their unique ecological environment is vulnerable to the impact of alien invasive plants. In recent years, island wetland vegetation has faced serious degradation problems, especially the large-scale invasion of Spartina alterniflora, which has posed a significant threat to the stability, biodiversity, and ecological functions of island ecosystems. The high-coverage invasion of Spartina alterniflora can form a single dominant community, displace native vegetation, and change wetland ecological processes. Therefore, accurately identifying and quantifying the coverage distribution of Spartina alterniflora is crucial for ecosystem management and invasive species prevention and control. Hyperspectral remote sensing technology provides an advanced tool for wetland vegetation monitoring due to its rich spectral information and high resolution. As an important means of remote sensing monitoring, vegetation indices can effectively evaluate the coverage and growth status of vegetation by quantifying the reflection characteristics of different spectral bands. However, different vegetation indices have different responses to vegetation coverage, especially in complex island environments, and their applicability has not been systematically evaluated. Therefore, determining the optimal vegetation index can provide a scientific basis for monitoring invasive plants in island ecosystems, improve monitoring accuracy and efficiency, and provide technical support for ecological restoration and management.

[0003] Currently, the application of vegetation indices in remote sensing monitoring mainly relies on common indices such as the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Soil Adjusted Vegetation Index (SAVI). These indices are widely used in vegetation growth monitoring, biomass estimation, and ecological assessment, but they have the following deficiencies in the island wetland environment: Soil and water interference: Conventional vegetation indices are easily affected by soil background and water reflection, resulting in large errors in coverage estimation. Especially in areas with low-coverage Spartina alterniflora distribution, the vegetation index may not be able to accurately distinguish vegetation from bare ground. Unclear applicability of different indices: Different vegetation indices have different sensitivities to Spartina alterniflora coverage, and there is a lack of systematic comparative analysis to screen the most suitable index for the island environment. Spatial scale adaptability: Current research is mostly based on medium- and low-resolution satellite data (such as Landsat, MODIS), while island ecosystems are usually small in area and complex in terrain, and traditional methods are difficult to meet the high-precision monitoring requirements. Lack of optimization methods combining hyperspectral data: Although hyperspectral remote sensing data provides rich spectral information, there is no unified method for screening the most suitable index for monitoring Spartina alterniflora coverage in islands among numerous vegetation indices.

[0004] Therefore, there is an urgent need for a method to overcome the problems of unclear applicability of traditional vegetation indices in island environments, large interference from soil and water bodies, and poor adaptability to spatial scales. By evaluating and analyzing various vegetation indices, the optimal vegetation index is selected to improve the recognition accuracy of Spartina alterniflora coverage. Summary of the Invention

[0005] In view of this, the present application provides a method and device for determining the optimal vegetation index of Spartina alterniflora, which are used to overcome the problems of unclear applicability of traditional vegetation indices in island environments, large interference from soil and water bodies, and poor adaptability to spatial scales. By evaluating and analyzing various vegetation indices, the optimal vegetation index is selected to improve the recognition accuracy of Spartina alterniflora coverage.

[0006] Specifically, the present application is implemented through the following technical solutions:

[0007] The first aspect of the present application provides a method for determining the optimal vegetation index of Spartina alterniflora, and the method includes:

[0008] Obtain hyperspectral image data of Spartina alterniflora; the hyperspectral image data includes continuous spectral data of Spartina alterniflora at multiple observation points, and the hyperspectral image data of each observation point includes reflectance information of multiple bands;

[0009] Based on the spectral characteristics of the hyperspectral image data and combining the spectral characteristics of various vegetation indices, determine the target band corresponding to each vegetation index; the target bands corresponding to different vegetation indices are different;

[0010] For each observation point, calculate each vegetation index at the observation point based on the reflectance information in the target band.

[0011] Obtain field sampling data of Spartina alterniflora; the field sampling data includes vegetation coverage data of Spartina alterniflora at multiple field sampling points;

[0012] For each observation point, match the vegetation index at the observation point with the vegetation coverage data at the corresponding field sampling point to establish a matching relationship;

[0013] For each vegetation index, calculate the correlation between the vegetation index at each observation point and the vegetation coverage at the corresponding field sampling point based on the matching relationship, and screen the optimal vegetation index based on the correlation.

[0014] The second aspect of the present application provides a device for determining the optimal vegetation index of Spartina alterniflora, and the device includes an acquisition module, a determination module, a calculation module, a matching module, and a screening module;

[0015] Among them, the acquisition module is used to acquire the hyperspectral image data of Spartina alterniflora; the hyperspectral image data includes the continuous spectral data of Spartina alterniflora under multiple observation points, and the hyperspectral image data of each observation point includes the reflectance information of multiple bands;

[0016] The determination module is used to determine the target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and in combination with the spectral characteristics of multiple vegetation indices;

[0017] The calculation module is used to calculate each vegetation index under each observation point respectively based on the reflectance information under the target band for each observation point;

[0018] The acquisition module is further used to acquire the field sampling data of Spartina alterniflora; the field sampling data includes the vegetation coverage data of Spartina alterniflora under multiple field sampling points;

[0019] The matching module is used to match the vegetation index under each observation point with the vegetation coverage data under the corresponding field sampling point for each observation point to establish a matching relationship;

[0020] The screening module is used to calculate the correlation between the vegetation index under each observation point and the vegetation coverage under the corresponding field sampling point based on the matching relationship for each vegetation index, and screen the optimal vegetation index based on the correlation.

[0021] The method and device for determining the optimal vegetation index of Spartina alterniflora provided by this application first utilize the spectral characteristics of hyperspectral image data to calculate different vegetation indices in order to obtain a preliminary estimated value of the coverage of Spartina alterniflora. The hyperspectral image data provides continuous band reflectance information, enabling the selection of the most suitable target bands according to the spectral characteristics of different vegetation indices to improve the effectiveness of index calculation. Then, based on the hyperspectral image data, the vegetation indices of each observation point are calculated to initially reflect the distribution and coverage of Spartina alterniflora. To improve the recognition accuracy of Spartina alterniflora, the estimated coverage value calculated from remote sensing data is matched with the field sampling data to verify and correct the effectiveness of the index. By matching the geographical coordinate information of the hyperspectral observation points and the field sampling points, a spatial mapping relationship is established, enabling the vegetation index of each observation point to be compared with the corresponding measured vegetation coverage. This process ensures the relevance between remote sensing data and ground observation data and improves the accuracy of subsequent analysis. After the matching relationship is established, the applicability of the vegetation index is evaluated by calculating the correlation between each vegetation index and the measured coverage, and the vegetation index with the highest correlation, the smallest error, and the most sensitivity to coverage changes is selected as the optimal vegetation index. The design of the entire process fully considers the combination of spectral data and measured data, avoiding the limitations that may be brought by a single vegetation index, making the selected vegetation index not only theoretically applicable to the monitoring of Spartina alterniflora, but also showing high accuracy in actual observations. By first calculating the preliminary estimated value of the coverage and then comparing and screening with the measured data, the entire vegetation index optimization process is scientific and data-driven, thereby improving the recognition accuracy of Spartina alterniflora and ensuring the accurate monitoring of the growth status and ecological changes of Spartina alterniflora under different environmental conditions. Description of the Drawings

[0022] Figure 1 It is a flowchart of the method for determining the optimal vegetation index of Spartina alterniflora provided by Embodiment 1 of this application;

[0023] Figure 2 It is a schematic structural diagram of the device for determining the optimal vegetation index of Spartina alterniflora provided by Embodiment 2 of this application. Detailed Embodiments

[0024] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0025] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0026] It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0027] Specific embodiments are given below to introduce the technical solutions of this application in detail.

[0028] Figure 1 It is a flowchart of the method for determining the optimal vegetation index of Spartina alterniflora provided in the first embodiment of this application.

[0029] Please refer to Figure 1 , the method provided in this embodiment may include:

[0030] S101. Obtain the hyperspectral image data of Spartina alterniflora.

[0031] Specifically, Spartina alterniflora is a perennial herbaceous plant. Due to its strong adaptability and reproductive ability, it has been introduced into many regions for coastal protection and wetland restoration. However, due to its extremely strong spreading ability and ecological competitiveness, Spartina alterniflora has become a globally recognized invasive species, causing serious impacts on coastal and wetland ecosystems. Due to the huge impact of Spartina alterniflora on wetland ecosystems, accurately monitoring its distribution and coverage changes is of great significance for ecological protection and the management of invasive species.

[0032] Furthermore, the hyperspectral image data includes the continuous spectral data of Spartina alterniflora at multiple observation points, and the hyperspectral image data of each observation point includes the reflectance information of multiple bands. The hyperspectral image data refers to the remote sensing data obtained through hyperspectral imaging technology, which contains the reflectance information of the ground object in multiple continuous spectral bands. The hyperspectral imaging system can record the spectral information in the range from visible light to near-infrared and even short-wave infrared with extremely high spectral resolution (nanometer level).

[0033] In specific implementation, the acquisition of the hyperspectral image data of Spartina alterniflora includes: determining the optimal sampling time based on the influence of seasonal changes and weather conditions on the growth state of Spartina alterniflora; determining the flight mode and flight altitude of the unmanned aerial vehicle (UAV) based on the influence of the UAV flight altitude on the spatial resolution and coverage; at the optimal sampling time, the UAV samples according to the preset grid route based on the flight mode and flight altitude to obtain hyperspectral image data; and performing geometric correction on the hyperspectral image data based on the three-dimensional terrain information obtained from multiple ground control points, and aligning the corrected hyperspectral image data with the true geographic coordinates.

[0034] Specifically, by analyzing the growth state of Spartina alterniflora in different seasons, a time period with the most obvious vegetation spectral characteristics and stable coverage (usually its vigorous growth period) is selected. Considering the influence of weather conditions on spectral image sampling, the selected time period is screened, and a time window with clear sky, less cloud cover, and lower wind speed is chosen to ensure stable and interference-free spectral data obtained by the hyperspectral sensor. According to the spatial resolution requirements of the hyperspectral sensor, the flight altitude of the UAV is adjusted to ensure that the collected data can guarantee clarity and cover a large enough area. Based on the endurance of the UAV and the task requirements, the flight path and flight mode of the UAV are set, including parameters such as flight line interval, overlap degree, and flight speed, to ensure full coverage and data continuity. At the determined optimal sampling time, the UAV is controlled to perform automatic aerial survey according to the preset grid route. A hyperspectral sensor is used to obtain hyperspectral images covering the research area, and the data is stored in real time, while the position information of the UAV (GPS coordinates, attitude angles, etc.) is recorded. The flight process of the UAV is monitored through ground station software and adjusted when necessary to ensure the integrity and quality of data collection. During the acquisition of hyperspectral image data, a coordinate reference system is established using the three-dimensional terrain information obtained from ground control points (GCPs) to provide accurate geographic positioning for the image data. Image registration techniques (such as nearest neighbor interpolation, bilinear interpolation, or resampling methods) are used to perform geometric correction on the collected hyperspectral image data to eliminate geometric distortions caused by changes in the flight attitude of the UAV or terrain undulations. Through geographic registration, the corrected hyperspectral image data is aligned with the true geographic coordinates for subsequent analysis and application.

[0035] For example, in this embodiment, a drone is used to carry a multi-sensor system to collect hyperspectral image data. The DJI drone is equipped with a lidar camera and a hyperspectral camera. The working wavelength range of the hyperspectral camera is from 400 to 1000 nanometers, covering the visible to near-infrared spectral regions. The spatial dimension resolution is 1920 pixels, and the number of spectral channels is as high as 1200, which can capture the fine spectral characteristics of ground objects. The lidar camera is used to obtain high-precision three-dimensional terrain and vegetation structure information, providing support for the spatial correction and analysis of hyperspectral image data. The sampling time is selected under clear and cloudless conditions in May, June, July-August, and September-October 2024 to ensure that the data quality is not interfered by the weather. Each flight of the drone adopts a low-altitude control flight mode, and the flight altitude is strictly controlled at 100 meters to ensure high spatial resolution and spectral accuracy of the data. The spatial resolution of the obtained hyperspectral image data is approximately 2 cm. The range of the data collection area is approximately 0.4 km 2 . During the flight, the drone conducts full-coverage sampling according to the preset grid-shaped flight path, and at the same time, geometric correction is performed in combination with ground control points (GCPs) to ensure the spatial consistency and accuracy of the data.

[0036] The method provided in the present embodiment, firstly, firstly, based on seasonal changes and weather conditions, the optimal sampling time is selected, which can ensure that the spectral characteristics of Spartina alterniflora are clear and stable, which helps to capture its typical growth state and minimize the interference of environmental factors such as clouds and shadows on spectral data. This precise time selection improves data quality and makes subsequent analysis more reliable. Secondly, by adjusting the flight altitude of the drone to optimize the spatial resolution and coverage, the accuracy of the hyperspectral image can be guaranteed, and the data of a large area can be efficiently obtained, taking into account local details and overall distribution. In addition, the gridded route is used for full coverage sampling to ensure the systematicness and consistency of data collection, avoid the data loss caused by the blind area of aerial photography, and improve the data integrity and repeatability. At the same time, the three-dimensional terrain information obtained by the ground control point is used for geometric correction, so that the hyperspectral image data is aligned with the real geographic coordinates, eliminating the geometric distortion caused by the change of the drone attitude and the undulation of the terrain, improving the spatial accuracy of the data, and providing accurate geographic reference for subsequent analysis. In the second aspect, when the optimal vegetation index is determined based on the hyperspectral image data in the follow-up, this data acquisition method plays a key role. The high precision and high consistency of hyperspectral image data make the calculation results of different vegetation indices more comparable, avoiding inaccurate analysis caused by data errors. Accurate spatial registration helps to correspond spectral information with the actual situation of the ground objects, making the calculation of vegetation indices based on spectral characteristics more physically meaningful and improving the reliability of the sensitivity assessment of different vegetation indices to cover changes. At the same time, the continuous spectral characteristics of hyperspectral data can support the calculation of richer vegetation indices, and can combine sensitivity analysis of different bands to screen out the index with the strongest response and the highest goodness of fit under the change of Spartina alterniflora cover. This data collection method not only improves the scientificity and accuracy of the analysis, but also provides a solid data foundation for the monitoring and management of Spartina alterniflora on islands.

[0037] S102: Determine a target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and in combination with the spectral characteristics of a plurality of vegetation indices.

[0038] Specifically, spectral characteristics refer to the reflection, absorption or scattering characteristics of an object in different electromagnetic wave bands. Different land objects (such as water bodies, soil, and vegetation) have different spectral reflectances in different wavelength ranges, thus forming unique spectral curves. For example, healthy vegetation has strong absorption in the red light (600-700nm) band and strong reflection in the near-infrared (700-1300nm) band. This typical "red edge" feature can be used to distinguish vegetation from other land objects and further analyze the growth status of vegetation.

[0039] Vegetation indices refer to mathematical combinations calculated using remote sensing data, usually composed of spectral reflectances at different wavelengths, and are mainly used to evaluate the growth status, health condition, coverage, and biomass of vegetation, etc. Its basic principle is to utilize the spectral reflectance characteristics of vegetation in the visible light (such as red light) and near-infrared light bands to calculate an index that can quantitatively reflect vegetation characteristics. There are various vegetation indices. In this embodiment, multiple vegetation indices include the Normalized Difference Vegetation Index (NDVI), the Soil Adjusted Vegetation Index (SAVI), the Green Normalized Difference Vegetation Index (GNDVI), the Enhanced Vegetation Index (EVI), the Weighted Atmosphere Vegetation Index (WAVI), and the Coastal Salt Marsh Vegetation Index (CSMVI). The calculation methods and principles of each vegetation index will be introduced in subsequent embodiments.

[0040] Furthermore, calculating vegetation indices can quantitatively evaluate the ecological characteristics of vegetation, facilitating the monitoring of its growth status, health condition, and coverage changes. Specifically, the calculation of vegetation indices can: enhance vegetation signals: Compared with the reflectance of a single band, vegetation indices can more effectively highlight vegetation information and reduce the interference of background noise on vegetation recognition. Distinguish vegetation with different coverages: Vegetation indices can reflect vegetation density and health, and play an important role in distinguishing Spartina alterniflora with different coverages.

[0041] It should be noted that the target bands corresponding to different vegetation indices are different. Due to different calculation purposes of different vegetation indices, the bands used by different vegetation indices have different emphases and are optimized for different environmental characteristics. For example, the Normalized Difference Vegetation Index mainly utilizes the reflectance difference between the red light and near-infrared bands and is applicable to the health assessment of most vegetation. The Enhanced Vegetation Index additionally introduces the blue light band to reduce the influence of atmospheric scattering and soil background and is applicable to high-coverage areas. The Soil Adjusted Vegetation Index is applicable to areas with low vegetation coverage by adjusting the soil background effect.

[0042] When specifically implemented, determining the target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and combining the spectral characteristics of multiple vegetation indices includes: performing spectral analysis on the hyperspectral image data to extract the spectral characteristics of Spartina alterniflora; based on the spectral characteristics, screening out candidate bands related to the growth status of Spartina alterniflora; based on the vegetation index calculation formula, analyzing the sensitivity of each vegetation index to the candidate bands to determine the band range of each vegetation index; analyzing the matching degree between the band range and the calculation requirements of the vegetation index to determine the target band.

[0043] Specifically, hyperspectral image data is read, and radiometric calibration and atmospheric correction are performed to eliminate the influence of environmental factors. The spectral reflectance of Spartina alterniflora is extracted within the region of interest (ROI) for Spartina alterniflora, and its spectral curve is calculated. The average reflectance of the Spartina alterniflora spectral curve at different wavelengths is calculated, and its spectral characteristics are analyzed, including the red edge position, spectral peak, valley value, etc. Further, the spectral reflection characteristics of Spartina alterniflora within the entire hyperspectral range are calculated and compared with other vegetation or backgrounds (such as water bodies, bare soil). Bands with high discrimination for Spartina alterniflora and sensitivity to growth status are selected as candidate bands, avoiding redundant or low - contribution bands. According to the range of candidate bands, the changes in various vegetation indices under different band combinations are calculated using vegetation index calculation formulas (such as NDVI, EVI, SAVI, etc.). A sensitivity analysis method (such as correlation analysis, regression analysis) is used to analyze the response degree of each vegetation index to the candidate bands, and a band range with high sensitivity is selected. The calculation results of different vegetation indices within a specific band range are compared to screen out bands with strong discrimination ability for the growth status of Spartina alterniflora. Combining the sensitivity analysis results, the calculation requirements of each vegetation index are compared with the selected band range to determine the optimally matched band combination. A statistical analysis method (such as principal component analysis, mutual information analysis) is used to further optimize the target bands, remove redundant information, improve the calculation accuracy, determine the final target bands for calculating each vegetation index, and apply them to subsequent vegetation index calculation and analysis.

[0044] The method provided in this embodiment can effectively improve the accuracy and scientific nature of data processing by analyzing the spectral characteristics of hyperspectral image data and combining the spectral characteristics of multiple vegetation indices. First, based on the specific spectral characteristics of Spartina alterniflora, the key bands that can best reflect its growth status are selected, avoiding the use of irrelevant or redundant bands, and improving the calculation efficiency and analysis accuracy. In addition, using the vegetation index calculation formula to analyze the sensitivity of each vegetation index to the candidate bands helps to screen out the band range with the highest discrimination for Spartina alterniflora, ensuring the rationality of vegetation index calculation, so that the finally selected target bands can maximize the reflection of the growth status and coverage of Spartina alterniflora. Moreover, this process has an important impact on the subsequent screening of the optimal vegetation index. First, through the precise matching of the target bands, the effectiveness of each vegetation index in Spartina alterniflora monitoring can be guaranteed, and the comparability between different indices can be improved. Second, this method can screen out bands that are irrelevant to Spartina alterniflora or have redundant information, making the final vegetation index calculation more concise and efficient, thereby enhancing the ability to identify the distribution and coverage of Spartina alterniflora. In addition, by matching the band ranges of different vegetation indices, the applicability of each vegetation index under different growth states and environmental conditions can be evaluated, and finally the optimal vegetation index that can most stably and efficiently reflect the coverage of Spartina alterniflora can be selected, laying a foundation for accurately identifying the coverage of Spartina alterniflora.

[0045] S103. For each observation point, based on the reflectance information in the target band, calculate the vegetation indices at the observation point respectively.

[0046] Specifically, as described above, the hyperspectral image data contains multiple observation points. Since the geographical locations of different observation points are different, their reflectance information is also different. Therefore, there are multiple vegetation indices corresponding to each observation point. The multiple vegetation indices include the Normalized Difference Vegetation Index (NDVI), the Soil Adjusted Vegetation Index (SAVI), the Green Normalized Vegetation Index (GNDVI), the Enhanced Vegetation Index (EVI), the Weighted Atmospherically Resistant Vegetation Index (WAVI), and the Coastal Zone Salt Marsh Vegetation Index (CZSVI). In this embodiment, the calculation processes of each vegetation index will be introduced in turn.

[0047] The Normalized Difference Vegetation Index (NDVI) is a common and widely used vegetation index. It uses the reflectance difference between the near-infrared (NIR) and red (Red) bands to measure the growth status and health level of vegetation. Its basic principle is based on the absorption and reflection characteristics of vegetation in different spectral bands: green vegetation has strong absorption in the red band (Red, usually 600 - 700 nm) because chlorophyll absorbs most of the visible light for photosynthesis. Green vegetation has strong reflection in the near-infrared band (NIR, usually 700 - 1300 nm) because the internal cell structure of the leaves scatters near-infrared light. Based on this characteristic, the calculation formula of the normalized vegetation index is as follows:

[0048]

[0049] where the NDVI is the Normalized Difference Vegetation Index; the ρ NIR is the reflectance in the near-infrared band; the ρ Red is the reflectance in the red band.

[0050] It should be noted that the Normalized Difference Vegetation Index (NDVI) is particularly suitable for large-scale vegetation dynamic monitoring. It can clearly distinguish different ground cover types and effectively highlight vegetation. Through ratio processing, the NDVI can partially eliminate the influence of irradiance condition changes (atmospheric path radiation) related to the solar elevation angle, satellite observation angle, terrain, cloud shadows, and atmospheric conditions. Its normalization process reduces the impact of sensor calibration decay, from 10%-30% for single-band influence to 0%-6% for the NDVI. At the same time, it reduces the angular influence caused by bidirectional surface reflectance and atmospheric effects. The NDVI is closely related to vegetation parameters such as Leaf Area Index (LAI), green biomass, vegetation coverage, and photosynthesis, and can accurately reflect the growth status and health level of vegetation. When the vegetation coverage exceeds a certain threshold (such as 80%), the value of the NDVI will tend to saturate and become less sensitive to changes in the vegetation canopy. This means that in areas with high vegetation coverage, the NDVI may not accurately reflect the growth changes of vegetation. The NDVI is relatively sensitive to changes in the soil background, and the reflection characteristics of bare soil may cause the NDVI value in low-coverage areas to be overestimated or underestimated, which to a certain extent affects the accuracy of the NDVI in low-vegetation coverage areas.

[0051] The Soil Adjusted Vegetation Index (SAVI) introduces a soil adjustment factor on the basis of the NDVI. The soil adjustment factor is used to reduce the influence of the soil background on the calculation of the vegetation index and is particularly suitable for areas with low vegetation coverage or more bare soil. In areas with sparse vegetation, the NDVI is easily affected by the soil reflectance because the soil also reflects light in the red (Red) and near-infrared (NIR) bands, resulting in distorted NDVI values. The SAVI adjusts the calculation formula so that the vegetation index can more stably reflect the vegetation status and provide more accurate vegetation information even under low coverage conditions. Based on this characteristic, the calculation formula of the SAVI is as follows:

[0052]

[0053] where SAVI is the Soil Adjusted Vegetation Index; L is the soil adjustment factor; ρ NIR is the reflectance in the near-infrared band; ρ Red is the reflectance in the red band.

[0054] It should be noted that the soil adjustment factor can be adjusted according to the vegetation density, so that the soil-adjusted vegetation index can better reflect the vegetation condition under different vegetation densities. The determination of the soil adjustment factor value depends on the accurate information of the vegetation density, but it is difficult to accurately know the vegetation density in practical applications. Therefore, it is difficult to optimize the soil-adjusted vegetation index. While reducing the influence of the soil background, the soil-adjusted vegetation index may also lose some vegetation signals, resulting in a lower vegetation index.

[0055] The green normalized difference vegetation index (GNDVI) is a variant of the normalized difference vegetation index, mainly used to enhance the detection ability of chlorophyll content. Compared with the normalized difference vegetation index, the GNDVI replaces the red (Red) band with the green (Green) band, making it more sensitive to chlorophyll content and photosynthesis level. In the spectral reflectance characteristics of vegetation, healthy vegetation has strong reflectance in the near-infrared (NIR) band and strong absorption in the visible light. Among them, the change in the green (Green, wavelength about 500 - 600 nm) band can better reflect the change in chlorophyll content. Therefore, the GNDVI is applicable to the middle and late stages of crop growth and can accurately reflect the health status, photosynthesis intensity, and physiological state of crops. Based on this characteristic, the calculation formula of the soil-adjusted vegetation index is as follows:

[0056]

[0057] Wherein, the GNDVI is the green normalized difference vegetation index; the ρ NIR is the reflectance in the near-infrared band; the ρ Green is the reflectance in the green band.

[0058] It should be noted that the calculation of the GNDVI depends on the data in the green band and the near-infrared band. If the data quality of these two bands is poor or there is noise, it may affect the accuracy of the GNDVI. The GNDVI is more applicable to the late stage of vegetation development or the stage of dense canopy, and may not be effective for the monitoring of the initial stage of vegetation development or sparse vegetation.

[0059] The Enhanced Vegetation Index (EVI) is an improvement over the Normalized Difference Vegetation Index (NDVI). By introducing atmospheric and soil adjustment factors, it enhances the sensitivity to vegetation in high-biomass areas. Since the NDVI tends to saturate in regions with high vegetation cover (i.e., the change in the NDVI slows down, making it difficult to further distinguish differences in high-density vegetation), the EVI corrects for the influence of atmospheric aerosols through the Blue Band and incorporates a soil adjustment factor, enabling a more accurate reflection of vegetation biomass and productivity. The EVI is mainly used in areas with high vegetation cover such as forests and tropical rainforests and has important value in photosynthesis monitoring, ecosystem modeling, and agricultural applications. The calculation of the EVI relies on high-quality remote sensing data, especially data in the Blue Band. Poor data quality or the presence of noise may affect the accuracy of the EVI.

[0060] Optionally, the vegetation index includes the Enhanced Vegetation Index. For each observation point, based on the reflectance information in the target band, the respective vegetation indices at the observation point are calculated as follows: respectively extract the first reflectance in the first target band, the second reflectance in the second target band, and the third reflectance in the third target band; calculate the first difference between the first reflectance and the second reflectance; calculate the first product of the first aerosol correction coefficient and the second reflectance, and calculate the second product of the second aerosol correction coefficient and the third reflectance; calculate the difference between the first reflectance and the first and second products, and calculate the first sum value of the difference and the soil adjustment factor; calculate the product of the gain factor and the first difference, and determine the quotient of the product and the first sum value as the Enhanced Vegetation Index.

[0061] Specifically, the first target band is the near-infrared band, the second target band is the red band, and the third target band is the blue band. The first aerosol correction coefficient is used to adjust the influence of the red band, and its typical value is 6.0. The first aerosol correction coefficient is mainly used to reduce the influence of atmospheric scattering on the red reflectance and ensure the calculation accuracy of the Enhanced Vegetation Index. The second aerosol correction coefficient is used to adjust the influence of the blue band, and its typical value is 7.0. Since the blue light is greatly affected by atmospheric aerosol scattering, the second aerosol correction coefficient helps with atmospheric correction in the calculation of the Enhanced Vegetation Index. The soil adjustment factor is used to reduce the influence of the soil background on the calculation of the Enhanced Vegetation Index, and its typical value is 1.0. Since the reflectance of bare soil or sparse vegetation areas is relatively high, the soil adjustment factor can reduce soil background interference and make the calculation of the Enhanced Vegetation Index more stable. The gain factor is used to enhance the contrast in the calculation of the vegetation index, and its typical value is 2.5. The gain factor magnifies the change in the Enhanced Vegetation Index value, increasing its sensitivity to vegetation biomass and enabling the Enhanced Vegetation Index to better monitor the vegetation status in high-biomass areas.

[0062] Based on the above characteristics, the calculation formula of the Enhanced Vegetation Index is as follows:

[0063]

[0064] Wherein, the EVI is the Enhanced Vegetation Index; the G is the gain factor; the ρ NIR is the reflectance in the near-infrared band; the ρ Red is the reflectance in the red light band; the ρ Blue is the reflectance in the blue light band; the L is the soil adjustment factor; the C1 is the first aerosol correction coefficient; the C2 is the second aerosol correction coefficient.

[0065] The Weighted Atmospheric Vegetation Index takes into account the influence of the atmosphere on the vegetation index during the calculation process. Through weighted processing, it can reduce the interference of atmospheric effects on vegetation monitoring to a certain extent and improve the accuracy of the vegetation index. The Weighted Atmospheric Vegetation Index has strong environmental adaptability and can provide relatively stable vegetation information under different atmospheric conditions and vegetation coverage. The calculation process of the Weighted Atmospheric Vegetation Index is relatively complex, involving weighted processing of multiple parameters and bands, and requires high computing power and data processing technology. The calculation of the Weighted Atmospheric Vegetation Index depends on high-quality remote sensing data. If the data quality is poor or there is noise, it may affect its accuracy and reliability.

[0066] Optionally, the vegetation index includes the Weighted Atmospheric Vegetation Index. For each observation point, based on the reflectance information in the target band, calculate each vegetation index under the observation point respectively, including: extract the first reflectance in the first target band and the third reflectance in the third target band respectively; calculate the second difference between the first reflectance and the third reflectance; calculate the second sum value of the first reflectance, the third reflectance, and the soil adjustment factor; calculate the product of the soil compensation factor and the second difference, and determine the quotient value of the product and the second sum value as the Weighted Atmospheric Vegetation Index. The calculation formula of the Weighted Atmospheric Vegetation Index is as follows:

[0067]

[0068] Wherein, the WAVI is the Weighted Atmospheric Vegetation Index; the ρ NIR is the reflectance in the near-infrared band; the ρ Blue is the reflectance in the blue light band; the L is the soil adjustment factor.

[0069] The coastal salt marsh vegetation index is a new remote sensing index specifically for salt marsh vegetation. By combining spectral characteristics and environmental factors, it can more accurately monitor and evaluate the growth status and ecological functions of salt marsh vegetation. It is constructed based on the reflection characteristics of salt marsh vegetation, in combination with vegetation indices and humidity indices. The core idea is to reduce the interference of high salinity and tidal changes on the calculation of vegetation indices by optimizing the band combination or introducing environmental adjustment factors. By comparison, an atmospheric-resistant vegetation index with good indication effect on vegetation coverage is selected, and the underlying surface soil moisture and water bodies are corrected by the normalized difference water index. It has important application value in the management and protection of coastal ecosystems.

[0070] Optionally, the vegetation index includes the coastal salt marsh vegetation index. For each observation point, based on the reflectance information in the target bands, the respective vegetation indices at the observation point are calculated as follows: The first reflectance in the first target band, the second reflectance in the second target band, the third reflectance in the third target band, and the fourth reflectance in the fourth target band are respectively extracted; the third product of the third aerosol correction coefficient and the second reflectance is calculated, and the third difference between the third product and the third reflectance is calculated; the difference between the first reflectance and the third difference is calculated, and the sum of the first reflectance and the third difference is calculated, and the quotient of the difference and the sum is determined as the first intermediate value; the difference between the fourth reflectance and the first reflectance is calculated, and the sum of the fourth reflectance and the first reflectance is calculated, and the quotient of the difference and the sum is determined as the second intermediate value; the sum of the soil compensation factor and the second intermediate value is calculated and determined as the third intermediate value; the product of the first intermediate value and the third intermediate value is determined as the coastal salt marsh vegetation index.

[0071] Specifically, the fourth target band is the green light band, and the third aerosol correction coefficient is used to adjust the influence of the red light band, and its usual value is 2.0. The calculation formula of the coastal salt marsh vegetation index is as follows:

[0072]

[0073] where, the CSMVI is the coastal salt marsh vegetation index; the ρ NIR is the reflectance in the near-infrared band; the ρ Red is the reflectance in the red light band; the ρ Blue is the reflectance in the blue light band; the L is the soil adjustment factor; the ρ GREEN is the reflectance in the green light band; the C3 is the third aerosol correction coefficient.

[0074] S104. Obtain the field sampling data of Spartina alterniflora.

[0075] Specifically, field sampling data includes the vegetation coverage data of Spartina alterniflora under multiple field sampling points, and field sampling data refers to the ecological environment data directly obtained by field investigation and measurement method. For Spartina alterniflora, field sampling data mainly includes information such as its growth state, biomass, vegetation coverage, chlorophyll content, soil moisture, etc. In this application, field sampling data pays special attention to vegetation coverage, i.e. the surface coverage ratio of Spartina alterniflora at sampling points. Vegetation coverage is usually expressed in percentage, for example: low coverage (<30%): sparse distribution, more exposed surface. Medium coverage (30%-70%): Spartina alterniflora grows more evenly, but there are still some exposed areas. High coverage (>70%): Spartina alterniflora is densely covered, and the surface is almost completely covered by vegetation.

[0076] In the specific implementation, based on the hyperspectral image data, combined with GIS (geographic information system), the typical growth area of Spartina alterniflora is selected. Multiple field sampling points are arranged by random sampling or systematic sampling to ensure that the field sampling points can cover different vegetation cover levels (low, medium and high coverage). A fixed area sample plot is set at each field sampling point to measure the vegetation cover of Spartina alterniflora. The coverage rate of Spartina alterniflora in the sample plot is calculated (expressed as a percentage) using visual estimation or digital image processing methods. Vegetation images are taken above the field sampling points using drones equipped with RGB, near-infrared or hyperspectral cameras. Environmental data related to the growth of Spartina alterniflora, such as soil moisture, salinity, water level, etc., are collected to facilitate the subsequent analysis of the factors affecting vegetation cover and record the vegetation cover data of each field sampling point.

[0077] For example, in the present embodiment, when obtaining field sampling data, researchers conducted field survey sampling on Xiaogan Island. The sampling time period will match the period between astronomical tide level 0.80m and 1.43m to ensure that the seawater conditions are suitable during data collection. The area where Spartina alterniflora is more concentrated on the island is selected as the survey sample area. These sample areas cover different vegetation states from low coverage to high coverage to ensure the representativeness and comprehensiveness of the data. A systematic sampling method was adopted in the sample area, and a total of 10 field sampling points were selected for detailed investigation. The area of each sample point is 10 meters × 10 meters, and the center coordinates of the sample point are accurately located by GPS. In the sample point, the coverage of Spartina alterniflora was recorded in grades: low coverage (<30%), medium coverage (30%-60%) and high coverage (>60%). Through visual estimation and sample box assistance, the coverage of Spartina alterniflora in each field sampling point was recorded in detail, and high-resolution photos were taken for subsequent analysis. In addition, a handheld ground object spectrometer was used to collect spectral reflectance, with a measured wavelength range of 200-1100nm. The azimuth angle of the spectrometer was 25°, and the spectral scan was set to 6 times, that is, each field sampling point was measured 6 times, and a whiteboard calibration was performed each time a different type of object was measured.

[0078] S105. For each observation point, match the vegetation index under the observation point with the vegetation cover data under the corresponding field sampling point to establish a matching relationship.

[0079] Specifically, when implementing, the step of for each observation point, matching the vegetation index under the observation point with the vegetation cover data under the corresponding field sampling point to establish a matching relationship includes: determining the spatial correspondence between the observation point and the field sampling point based on the coordinate information in the hyperspectral image data and the geographical location data of the field sampling point; establishing a spatial mapping between the observation point and the field sampling point based on the spatial correspondence; for each observation point, matching the vegetation index under the observation point with the vegetation cover data under the corresponding field sampling point.

[0080] Specifically, extract the geographical coordinate information (such as longitude and latitude, projection coordinates) in the hyperspectral image data. Collect the geographical location information of the field sampling points (record coordinates using a GPS device). Determine the spatial correspondence between the observation point and the field sampling point through spatial interpolation or the nearest neighbor method. Based on the determined spatial correspondence, construct a spatial mapping relationship between the observation point and the field sampling point. If there is an overlapping area between the observation point and multiple field sampling points, a weighted average or interpolation method can be used to determine the best match. According to the spatial mapping, match the vegetation index corresponding to the observation point with the vegetation cover data of the field sampling point one by one, and store the matched data to provide a basic data set for subsequent analysis.

[0081] The method provided in this embodiment helps to improve the correspondence between the vegetation index and the vegetation cover by establishing a matching relationship between the observation point and the field sampling point, providing a reliable reference basis for subsequent screening of the optimal vegetation index. By determining the spatial correspondence, it ensures the spatial consistency between the hyperspectral data and the field sampling data, avoids data mis-matching caused by geographical deviation, and improves the accuracy of vegetation index calculation. The matching process enables the vegetation index of each observation point to be directly compared with the vegetation cover measured in the field, thereby evaluating the sensitivity and applicability of different vegetation indices to the vegetation cover. This can quantify the correlation between each vegetation index and the actual vegetation cover, and further identify the optimal vegetation index that can most accurately reflect the growth status of Spartina alterniflora. In addition, by establishing a data mapping through matching, errors can be analyzed and corrected, improving the accuracy of Spartina alterniflora identification and providing higher-precision support for subsequent vegetation monitoring and ecological assessment.

[0082] S106. For each vegetation index, calculate the correlation between the vegetation index under each observation point and the vegetation cover under the corresponding field sampling point based on the matching relationship, and screen the optimal vegetation index based on the correlation.

[0083] In specific implementation, for each vegetation index, calculate the correlation between the vegetation index at each observation point and the vegetation coverage at the corresponding field sampling point based on the matching relationship, and screen the optimal vegetation index based on the correlation, including: for the observation points and the corresponding field sampling points in each group, calculate the correlation between the vegetation index and the vegetation coverage in the group, and screen the first vegetation index based on the correlation; substitute the first vegetation index into all the other groups, calculate the overall correlation of the first vegetation index in all groups, and screen the optimal vegetation index based on the overall correlation.

[0084] Specifically, for each group (i.e., different location areas or environmental conditions), extract the vegetation index values of all observation points in the group and the vegetation coverage values of their corresponding field sampling points. Use a correlation analysis method (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) to calculate the correlation between each vegetation index and the vegetation coverage. Select the vegetation index with the highest correlation in the group as the first vegetation index. Further, substitute the first vegetation index into all the other groups, and recalculate the correlation of the observation points and field sampling points in each group under all groups. Statistically analyze the overall correlation of the first vegetation index in all groups to determine its stability and applicability under different environmental conditions. Compare the overall correlations of all the first vegetation indices and select the first vegetation index with the best performance in all groups as the final optimal vegetation index.

[0085] Optionally, the step of calculating the correlation between the vegetation index and the vegetation coverage in the group for the observation points and the corresponding field sampling points in each group and screening the first vegetation index based on the correlation includes: for each observation point and the corresponding field sampling point, establish a matching relationship between the vegetation index and the vegetation coverage, and construct a fitting curve through regression analysis; calculate the determination coefficient, root mean square error, and slope of the fitting curve; the slope characterizes the sensitivity of the vegetation index to the change in vegetation coverage; for each observation point, calculate the comprehensive evaluation value of each vegetation index through weighted summation based on the weights of the determination coefficient, root mean square error, and slope, and determine the vegetation index with the largest comprehensive evaluation value as the first vegetation index of the observation point.

[0086] In specific implementation, for each observation point, obtain the vegetation coverage data of its corresponding field sampling point. Through matching, construct a dataset of the vegetation index and the vegetation coverage. Use a regression analysis method (such as linear regression or non - linear regression), with the vegetation coverage as the dependent variable and the vegetation index as the independent variable, to fit the relationship curve between the vegetation index and the vegetation coverage. Calculate the determination coefficient (R 2), to measure the explanatory power of the vegetation index for the change in vegetation coverage. Calculate the root mean square error (RMSE) of the fitting curve to measure the magnitude of the fitting error. Calculate the slope (S) of the fitting curve to characterize the sensitivity of the vegetation index to the change in vegetation coverage. Set corresponding weights for the coefficient of determination, root mean square error, and slope, and calculate the comprehensive evaluation value of each vegetation index through weighted summation:

[0087] V = w1·R 2 - w2·RMSE + w3·S;

[0088] where, the V is the comprehensive evaluation value; the w1 is the weight of the coefficient of determination; the R 2 is the coefficient of determination; the w2 is the weight of the root mean square error; the RMSE is the root mean square error; the w3 is the weight of the slope; the S is the slope.

[0089] Select the vegetation index with the largest comprehensive evaluation value as the first vegetation index at this observation point.

[0090] When setting the weights of the coefficient of determination, root mean square error, and slope, it can be determined in various ways. The empirical weighting method is based on research objectives and experience to manually set weights. For example: The coefficient of determination reflects the explanatory power of the vegetation index for the change in vegetation coverage and is generally the most important, so a relatively high weight (such as 0.5) can be assigned. The root mean square error measures the fitting error, and the smaller the error, the better, so a negative weight (such as -0.3) is assigned. The slope characterizes the sensitivity of the vegetation index, and a larger value represents a more obvious response of the vegetation index to the change in vegetation coverage, and a positive weight (such as 0.2) is generally assigned. The analytic hierarchy process (AHP) compares the importance of the three through expert scoring to construct a pairwise comparison matrix. By calculating the eigenvalues and weights, ensure that the consistency test passes. The data-driven method (regression optimization method) optimizes the weights through a large amount of sample data using multiple regression or machine learning methods (such as ridge regression, Lasso regression). The goal is to maximize the model fitting degree or minimize the error so that the weights can automatically adapt to the data characteristics. The entropy weight method (based on data variability) calculates the information entropy of each parameter. The larger the information entropy, the lower the importance of the parameter. Weights are assigned based on the information entropy to reduce human intervention. The cross-validation method uses different weight combinations for cross-validation and selects the weight combination with the best performance on the test set.

[0091] In the method provided by this embodiment, in the first aspect, first, using the spectral characteristics of hyperspectral image data, different vegetation indices are calculated to obtain a preliminary estimated value of the Spartina alterniflora coverage. The hyperspectral image data provides continuous band reflectance information, enabling the selection of the most suitable target bands according to the spectral characteristics of different vegetation indices to improve the effectiveness of index calculation. Then, based on the hyperspectral image data, the vegetation indices of each observation point are calculated to initially reflect the distribution and coverage of Spartina alterniflora. To improve the accuracy of Spartina alterniflora identification, the coverage estimated value calculated from remote sensing data is matched with the field sampling data to verify and correct the effectiveness of the index. By matching the geographical coordinate information of hyperspectral observation points and field sampling points, a spatial mapping relationship is established, enabling the vegetation index of each observation point to be compared with the corresponding measured vegetation coverage. This process ensures the relevance between remote sensing data and ground observation data and improves the accuracy of subsequent analysis. After the matching relationship is established, the applicability of the vegetation index is evaluated by calculating the correlation between each vegetation index and the measured coverage, and the vegetation index with the highest correlation, the smallest error, and the most sensitivity to coverage changes is selected as the optimal vegetation index. The design of the entire process fully considers the combination of spectral data and measured data, avoiding the limitations that may be brought by a single vegetation index, making the selected vegetation index not only theoretically applicable to the monitoring of Spartina alterniflora but also showing high accuracy in actual observations. By first calculating the preliminary estimated value of coverage and then comparing and screening with the measured data, the entire vegetation index optimization process is scientific and data-driven, thereby improving the identification accuracy of Spartina alterniflora and ensuring the accurate monitoring of the growth status and ecological changes of Spartina alterniflora under different environmental conditions. In the second aspect, through the spectral feature analysis of hyperspectral image data and combining the spectral characteristics of multiple vegetation indices, the target bands of each vegetation index are gradually determined, which can effectively improve the accuracy and scientificity of data processing. First, based on the specific spectral characteristics of Spartina alterniflora, the key bands that can best reflect its growth status are selected, avoiding the use of irrelevant or redundant bands, and improving the calculation efficiency and analysis accuracy. In addition, analyzing the sensitivity of each vegetation index to the candidate bands using the vegetation index calculation formula helps to screen out the band range with the most discrimination for Spartina alterniflora, ensuring the rationality of vegetation index calculation and enabling the finally selected target bands to maximize the reflection of the growth state and coverage of Spartina alterniflora. Moreover, this process has an important impact on the subsequent screening of the optimal vegetation index. First, through the accurate matching of the target bands, the effectiveness of each vegetation index in Spartina alterniflora monitoring can be guaranteed, and the comparability between different indices can be improved. Second, this method can screen out the bands that are irrelevant to Spartina alterniflora or have redundant information, making the final vegetation index calculation more concise and efficient, thereby enhancing the ability to identify the distribution and coverage of Spartina alterniflora.In addition, by matching the band ranges of different vegetation indices, the applicability of each vegetation index under different growth states and environmental conditions can be evaluated, and finally the optimal vegetation index that can most stably and efficiently reflect the coverage of Spartina alterniflora can be selected, laying a foundation for accurately identifying the coverage of Spartina alterniflora.

[0092] Corresponding to the foregoing embodiment of the method for determining the optimal vegetation index of Spartina alterniflora, the present application also provides an embodiment of a device for determining the optimal vegetation index of Spartina alterniflora.

[0093] Figure 2 It is a schematic structural diagram of the device for determining the optimal vegetation index of Spartina alterniflora provided in the second embodiment of the present application. Please refer to Figure 2 The device provided in this embodiment includes an acquisition module 210, a determination module 220, a calculation module 230, a matching module 240, and a screening module 250;

[0094] Among them, the acquisition module 210 is configured to acquire hyperspectral image data of Spartina alterniflora; the hyperspectral image data includes continuous spectral data of Spartina alterniflora under multiple observation points, and the hyperspectral image data of each observation point includes reflectance information of multiple bands;

[0095] The determination module 220 is configured to determine the target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and in combination with the spectral characteristics of multiple vegetation indices;

[0096] The calculation module 230 is configured to calculate each vegetation index at each observation point based on the reflectance information under the target band for each observation point;

[0097] The acquisition module 210 is further configured to acquire field sampling data of Spartina alterniflora; the field sampling data includes vegetation coverage data of Spartina alterniflora under multiple field sampling points;

[0098] The matching module 240 is configured to match the vegetation index at each observation point with the vegetation coverage data at the corresponding field sampling point for each observation point to establish a matching relationship;

[0099] The screening module 250 is configured to calculate the correlation between the vegetation index and the vegetation coverage at the corresponding field sampling point at each observation point for each vegetation index based on the matching relationship, and screen the optimal vegetation index based on the correlation.

[0100] The device in this embodiment can be used to execute Figure 1 the steps of the method embodiment shown, and the specific implementation principle and process are similar, which will not be elaborated here.

[0101] For the implementation processes of the functions and roles of the respective units in the above device, please refer to the implementation processes of the corresponding steps in the above method for details, which will not be elaborated herein.

[0102] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial descriptions of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0103] The above are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A method for determining the optimal vegetation index of Spartina alterniflora, characterized in that, The method includes: Obtaining hyperspectral image data of Spartina alterniflora; the hyperspectral image data includes continuous spectral data of Spartina alterniflora at multiple observation points, and the hyperspectral image data at each observation point includes reflectance information of multiple bands; Based on the spectral characteristics of the hyperspectral image data and combining the spectral characteristics of multiple vegetation indices, determining the target band corresponding to each vegetation index; the target bands corresponding to different vegetation indices are different; For each observation point, calculating each vegetation index at the observation point based on the reflectance information in the target band; Obtaining field sampling data of Spartina alterniflora; the field sampling data includes vegetation coverage data of Spartina alterniflora at multiple field sampling points; For each observation point, matching the vegetation index at the observation point with the vegetation coverage data at the corresponding field sampling point to establish a matching relationship; For each vegetation index, calculating the correlation between the vegetation index and the vegetation coverage at the corresponding field sampling point at each observation point based on the matching relationship, and screening the optimal vegetation index based on the correlation; 2. The method according to claim 1, wherein The determining the target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and combining the spectral characteristics of multiple vegetation indices includes: Performing spectral analysis on the hyperspectral image data to extract the spectral characteristics of Spartina alterniflora; Based on the spectral characteristics, screening out candidate bands related to the growth status of Spartina alterniflora; Based on the vegetation index calculation formula, analyzing the sensitivity of each vegetation index to the candidate bands to determine the band range of each vegetation index; Analyzing the matching degree between the band range and the vegetation index calculation requirements to determine the target band; 3. The method according to claim 1, characterized in that, The matching the vegetation index at each observation point with the vegetation coverage data at the corresponding field sampling point to establish a matching relationship includes: Based on the coordinate information in the hyperspectral image data and the geographical location data of the field sampling points, determining the spatial correspondence between the observation points and the field sampling points; Based on the spatial correspondence, establishing a spatial mapping between the observation points and the field sampling points; For each observation point, matching the vegetation index at the observation point with the vegetation coverage data at the corresponding field sampling point; 4. The method according to claim 1, characterized in that For each vegetation index, calculating the correlation between the vegetation index and the vegetation coverage at the corresponding field sampling point at each observation point based on the matching relationship, and screening the optimal vegetation index based on the correlation includes: For the observation points and the corresponding field sampling points in each group, calculating the correlation between the vegetation index and the vegetation coverage in the group, and screening the first vegetation index based on the correlation; Substituting the first vegetation index into all the other groups, calculating the overall correlation of the first vegetation index in all groups, and screening the optimal vegetation index based on the overall correlation; 5. The method according to claim 4, characterized in that, The calculating the correlation between the vegetation index and the vegetation coverage in the group for the observation points and the corresponding field sampling points in each group and screening the first vegetation index based on the correlation includes: For each observation point and its corresponding field sampling point, establish the matching relationship between the vegetation index and the vegetation cover, and construct a fitting curve through regression analysis; Calculate the determination coefficient, root mean square error, and slope of the fitting curve; the slope characterizes the sensitivity of the vegetation index to the change in vegetation cover; For each observation point, calculate the comprehensive evaluation value of each vegetation index through weighted summation based on the weights of the determination coefficient, root mean square error, and slope, and determine the vegetation index with the largest comprehensive evaluation value as the first vegetation index of the observation point.

6. The method according to claim 1, wherein The obtaining of the hyperspectral image data of Spartina alterniflora includes: Based on the influence of seasonal changes and weather conditions on the growth state of Spartina alterniflora, determine the optimal sampling time; Based on the influence of the UAV flight altitude on the spatial resolution and coverage, determine the flight mode and flight altitude of the UAV; At the optimal sampling time, the UAV samples according to the preset grid route based on the flight mode and flight altitude to obtain hyperspectral image data; Perform geometric correction on the hyperspectral image data based on the three-dimensional terrain information obtained from multiple ground control points, and align the corrected hyperspectral image data with the true geographical coordinates.

7. The method according to claim 1, wherein The vegetation index includes the Enhanced Vegetation Index. For each observation point, based on the reflectance information in the target band, calculate each vegetation index at the observation point respectively, including: Extract the first reflectance in the first target band, the second reflectance in the second target band, and the third reflectance in the third target band respectively; Calculate the first difference between the first reflectance and the second reflectance; Calculate the first product of the first aerosol correction coefficient and the second reflectance, and calculate the second product of the second aerosol correction coefficient and the third reflectance; Calculate the difference between the first reflectance and the first product and the second product, and calculate the first sum value of the difference and the soil adjustment factor; Calculate the product of the gain factor and the first difference, and determine the quotient of the product and the first sum value as the Enhanced Vegetation Index.

8. The method according to claim 1, characterized in that, The vegetation index includes the Weighted Atmospherically Resistant Vegetation Index. For each observation point, based on the reflectance information in the target band, calculate each vegetation index at the observation point respectively, including: Extract the first reflectance in the first target band and the third reflectance in the third target band respectively; Calculate the second difference between the first reflectance and the third reflectance; Calculate the second sum value of the first reflectance, the third reflectance, and the soil adjustment factor; Calculate the product of the soil compensation factor and the second difference, and determine the quotient of the product and the second sum value as the Weighted Atmospherically Resistant Vegetation Index.

9. The method according to claim 1, wherein The vegetation index includes the Coastal Zone Salt Marsh Vegetation Index. For each observation point, based on the reflectance information in the target band, calculate each vegetation index at the observation point respectively, including: Extract the first reflectance in the first target band, the second reflectance in the second target band, the third reflectance in the third target band, and the fourth reflectance in the fourth target band respectively; Calculate the third product of the third aerosol correction factor and the second reflectance, and calculate the third difference between the third product and the third reflectance; Calculate the difference between the first reflectance and the third difference, calculate the sum of the first reflectance and the third difference, and determine the quotient of the difference and the sum as the first intermediate value; Calculate the difference between the fourth reflectance and the first reflectance, calculate the sum of the fourth reflectance and the first reflectance, and determine the quotient of the difference and the sum as the second intermediate value; Calculate the sum of the soil compensation factor and the second intermediate value, and determine it as the third intermediate value; Determine the product of the first intermediate value and the third intermediate value as the coastal salt marsh vegetation index.

10. An apparatus for determining the optimal vegetation index of Spartina alterniflora, characterized in that, The device includes an acquisition module, a determination module, a calculation module, a matching module, and a screening module; Among them, the acquisition module is used to acquire the hyperspectral image data of Spartina alterniflora; the hyperspectral image data includes the continuous spectral data of Spartina alterniflora under multiple observation points, and the hyperspectral image data of each observation point includes the reflectance information of multiple bands; The determination module is used to determine the target band corresponding to each vegetation index based on the spectral characteristics of the hyperspectral image data and in combination with the spectral characteristics of multiple vegetation indices; The calculation module is used to calculate each vegetation index at each observation point based on the reflectance information in the target band for each observation point; The acquisition module is further used to acquire the field sampling data of Spartina alterniflora; the field sampling data includes the vegetation coverage data of Spartina alterniflora under multiple field sampling points; The matching module is used to match the vegetation index at each observation point with the vegetation coverage data at the corresponding field sampling point for each observation point to establish a matching relationship; The screening module is used to calculate the correlation between the vegetation index and the vegetation coverage at the corresponding field sampling point at each observation point based on the matching relationship for each vegetation index, and screen the optimal vegetation index based on the correlation.

Images