Sea ice condition impact factor evaluation method and system based on remote sensing

Through remote sensing technology, combined with grayscale symbiosis matrix and Pearson-III distribution model, the problem of low monitoring accuracy of sea ice distribution characteristics is solved, high-precision sea ice thickness and recurrence period prediction are achieved, and safety design of marine engineering is supported.

CN118982767AInactive Publication Date: 2024-11-19OCEAN UNIV OF CHINA

Patent Information

Application Number
CN202411468882.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-21
Publication Date
2024-11-19
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art has low accuracy in monitoring sea ice distribution characteristics of coastal sea areas, making it difficult to achieve high-precision sea ice thickness and recurrence period prediction.

Method used

The sea ice ice impact factor evaluation method based on remote sensing is used to extract the sea ice area time series through MODIS L1B data, and the sea ice texture features are extracted in combination with the gray symbiosis matrix operation of Sentinel-1A/B data. The narrow band wide band mutual conversion formula is used. The sea ice thickness inversion was performed by the model of the relationship between the wide band albedo and sea ice thickness, and the Pearson-III distribution model was used to calculate the spatial distribution of sea ice extreme thickness under different recurrence periods.

Benefits of technology

It improves the accuracy of sea ice monitoring, can quickly form sea ice range and thickness changes data, provides reliable sea ice design parameters, supports safety guarantees for navigation, aquaculture, and engineering facilities, and reduces risks in engineering design.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982767A_ABST
    Figure CN118982767A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of sea ice condition information processing, and discloses a remote sensing-based sea ice condition impact factor evaluation method and system. According to the method, MODIS L1B data of a certain sea area in winter for many years are selected, a sea ice area time sequence of the certain sea in winter for many years is extracted, sea ice texture features are extracted based on gray-level co-occurrence matrix operation by using Sentinel-1A / B data in winter for many years, sea ice types are identified, and then optical observation sea ice images are supplemented and improved. Performing inversion on the sea ice thickness of a certain sea to obtain a sea ice thickness and spatial distribution diagram of the certain sea in winter for many years, and analyzing a sea ice spatial distribution rule; obtaining the relationship between the air temperature, the sea surface temperature, the wind field and the salinity and the sea ice and the ice condition grade; based on the sea ice thickness inversion data, the Pearson-III distribution model is utilized to solve the sea ice extreme value thickness space distribution of a certain sea in different return periods, and the monitoring precision is high.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of sea ice information processing, and in particular relates to a method and system for evaluating sea ice influencing factors based on remote sensing. Background Art

[0002] Traditional means of sea ice monitoring mainly rely on establishing observation stations or radar stations along the coast and manual observation, combined with traditional methods such as ship monitoring and aircraft monitoring, which cannot achieve rapid monitoring over a large area. The advantages of remote sensing technology in observing a wide area and quickly obtaining ground information have greatly improved the efficiency of sea ice monitoring.

[0003] Optical remote sensing monitoring. Compared with traditional observation methods, remote sensing monitoring can obtain large-area sea ice information in real time at low cost. Sea ice monitoring based on optical satellite data mostly uses visible light bands and thermal infrared bands. The polar visibility conditions are poor, so SAR satellite monitoring is the main means of sea ice monitoring in polar regions. The meteorological conditions in a certain sea area are relatively good, and a sea ice monitoring method combining optical remote sensing and microwave remote sensing can be used. Shi et al. used MODIS-Aqua satellite observation data in a certain sea area in the winter of 2002-2010 to perform short-wave infrared atmospheric correction on the 1640nm and 2130nm bands, and found that snow accumulation in a certain sea area had little effect on sea ice optical products and SWIR atmospheric correction. Su et al. used gray-level co-occurrence matrix to extract the characteristic texture of MODIS 250m data, identified the sea ice range and ice edge line through texture segmentation method, and compared the MODIS data with the simultaneous and spatial HJ1B image to identify and classify sea ice. Yuan et al. proposed a semi-empirical model of sea ice thickness that can estimate the thickness of one year's ice using a water depth model and hyperspectral remote sensing data. Yu et al. obtained the sea ice area of ​​a certain sea from 1985 to 2015 based on multi-source optical remote sensing data (MODIS, AVHRR, GOCI, Landsat 1-5), established a piecewise multivariate nonlinear regression model, and combined meteorological data with sea ice area data to reconstruct a long-term daily sea ice area dataset for a certain sea from 1985 to 2015.

[0004] Since then, scholars have begun to use various optical remote sensing data to study sea ice in a certain sea. Zhu Yaping et al. used FY-2C meteorological satellite data to monitor sea ice in a certain sea, and proposed a basis for quantitative monitoring of sea ice based on geostationary meteorological satellites. Wang Zhiyong et al. used Landsat-7 / 8 and Sentinel-2 optical satellite data to explore the characteristic differences between sea ice and seawater spectra, and used the empirical inversion model of sea ice thickness to invert sea ice thickness.

[0005] According to the research on monitoring sea ice using various optical remote sensing satellite data, MODIS optical remote sensing data is the most widely used in sea ice monitoring due to its advantages such as high spatial resolution, short imaging period and free global download.

[0006] SAR satellite sea ice monitoring. Microwave remote sensing has a short history, but microwave remote sensing has stronger penetration than optical remote sensing, is less affected by complex weather, and can image at night, which can realize full-climate, all-day and large-scale monitoring of sea ice. Luo Licheng et al. used Sentinel-1 A / B satellite data to extract feature textures of McClure Strait sea ice to identify and classify sea ice. Shokr et al. proposed that in the first 0.3m of sea ice growth, all back-reflection parameters increase with the increase of sea ice thickness, while the two polarization decomposition parameters H and α decrease in the same range. In addition, through research, it was found that the reason why the dielectric constant pre-backscattering changes significantly when the sea ice thickness reaches 0.3 m is that a large amount of salinity reaches the drainage limit when the sea ice thickness is 0.3m. Zhao Quanhua et al. used Sentinel-1A / B data from a certain bay and sea area to compare the accuracy of extracting sea ice characteristic parameters using the gray-level co-occurrence matrix and the Bhattacharyya distance method in sea ice identification, and then selected the optimal texture parameter combination to classify the sea ice. They also studied the ice conditions in a certain bay in combination with the sea ice area and ice edge.

[0007] In summary, optical remote sensing data and SAR remote sensing data have their own advantages in sea ice monitoring. If you need a higher spatial resolution, you can choose Landsat data, high-resolution data, etc. If you need a long-term series of sea ice monitoring, you can choose MODIS satellite data with higher temporal resolution; SAR remote sensing images such as Sentinel-1A / B are not affected by bad weather and can obtain sea ice data in complex weather such as rain and snow. However, previous monitoring of sea ice in a certain sea was mostly based on a single satellite image, and the effective sea ice data obtained was limited.

[0008] Calculation of sea ice thickness return period. The sea ice return period refers to the frequency and time period of sea ice appearance in an ice-covered sea area for a certain period of time, indicating the periodicity of sea ice formation and melting in a specific area. Specifically, the sea ice return period indicates the time interval from the beginning of sea ice formation in one sea area to the next formation of sea ice. Sea ice thickness is an important indicator in sea ice information. Usually, sea ice in coastal areas will accumulate and stack under the action of ocean dynamics. Affected by factors such as terrain, the ice layer with a smooth surface is often composed of multiple overlapping layers. Therefore, the observation of sea ice with a smooth surface is challenging. In the long-term prediction of the marine environment, the specification recommends the use of Pearson-Ⅲ, Gumbel, Weibull or log-normal distribution and other calculation methods for recurrence value calculation. These methods are suitable for the calculation of different data volumes and time series, so it is necessary to consider the actual factors when choosing. Pearson-Ⅲ distribution can achieve a better fitting effect through the line fitting method or adjusting the deviation coefficient and mean, but the prediction accuracy is poor when the data time series is short. The Weibull distribution is flexible in application. The slope and intercept can be determined through the general diagram, and the parameters can also be obtained by the least squares method. The Gumbel distribution has a good fit to the empirical points near the median, but it has difficulties in handling very large values. The lognormal distribution avoids the trouble of selecting the fitting parameters, but requires a large amount of data to ensure the accuracy of the calculation.

[0009] When determining the design ice thickness, an estimate of the sea ice occurrence point is usually given. However, the accuracy of the point estimate will be biased by the differences in the distribution model and the estimation method. In order to reduce the bias, more accurate data and models can be used, and multiple estimation methods can be adopted and compared and verified. At the same time, it is also possible to consider using more refined grid division and simulation methods to comprehensively consider the influence of multiple factors (such as wind direction, sea water temperature, etc.) on sea ice to improve the accuracy of predicting the possibility and thickness of sea ice. Coles proposed a widely applicable interval estimation method for the quantile of extreme value distribution by using the maximum likelihood method. This method is suitable for setting parameters of ice thickness in hydrological assessment and can also be applied to other similar problems. Tao Shanshan et al. used the same method to determine the annual total ice volume measurement data in the northern part of a certain sea to find the credible interval of the parameters they designed, which further expanded the use of this method to solve other types of problems. The results show that in terms of the sea ice return period in the waters near Yingkou, the Pearson-Ⅲ distribution has the best accuracy. The smaller the return period, the higher the prediction accuracy. However, when the return period is large, an extended prediction is required, resulting in a decrease in prediction accuracy. Zhang Shude and others used satellite image information of sea ice in a certain bay over the years to deduce factors such as sea ice thickness and density. Then, based on the sea ice factors and the Gumbel method, they obtained the distribution of sea ice thickness in various recurrence periods in a certain bay and a sea area. This plays a key role in the anti-freezing design of marine engineering in a certain bay and the safety work in the winter sea area.

[0010] In terms of factors affecting sea ice. A certain sea is surrounded by land on three sides, and climate change on land has a great impact on it. Therefore, local meteorological factors such as average daily temperature have become direct factors affecting the ice conditions in a certain sea. Based on MODIS data, Li Yanqing analyzed the offshore average time series of the outer edge of the sea ice in a certain sea from 2001 to 2011 and various meteorological factors. The study found that the sea ice in small sea areas responds faster to temperature than that in large sea areas. Liu Cheng et al. used NSIDC sea ice products with high temporal and spatial resolution to analyze the spatiotemporal changes in the sea ice range of a certain sea from 2007 to 2018. The results showed that the sea ice range of a certain sea showed a trend of first increasing and then decreasing from 2007 to 2018, and the maximum sea ice range occurred from late January to late February. Sun Jingqi and others used optical remote sensing data such as AVHRR and MODIS to study the sea ice conditions in a certain sea from 1988 to 2018, and used correlation analysis to find that the meteorological factors most correlated with sea ice conditions were cumulative freezing degree days (CFDD) and cumulative melting degree days (CTDD); and selected CFDD, CTDD, 3d-1.8℃ accumulated temperature and minimum temperature to establish a sea ice simulation model for a certain sea.

[0011] Salinity is an important factor in seawater and a key factor in the formation of sea ice. Salt cannot be locked into the sea ice lattice, but is dissolved in the liquid content of brine. The salinity of ice varies in space and time due to the retention and drainage of brine. Salinity is related to temperature and controls the relative amount of brine enclosed in the ice, which in turn affects the properties of the ice. Currently, there are few studies on the relationship between salinity and sea ice, and they are concentrated in polar regions. Martin took the salinity factor into account in the sea ice model simulating the Arctic and Antarctic to make the model predictions more realistic. Wang Qingkai studied the thermodynamic effects of salinity differences on sea ice thickness and sea ice density and showed that during the freezing process of sea ice, the salinity difference can increase the temperature of sea ice and inhibit freezing.

[0012] In terms of the impact of sea ice on the environment. Previous studies have shown that the coverage of sea ice will affect the waves and ocean currents in the covered sea area and the sediment resuspension caused. Most of the research on wave propagation under the influence of sea ice focuses on the sea areas with developed sea ice at high latitudes. However, in recent years, more scholars have paid attention to the impact of mid-latitude sea ice on local hydrodynamics and sediments. Yue et al. found that the average effective wave height of a certain sea was attenuated by 16.52% during the ice-covered period. Zhang et al. studied the wave energy flux in a certain bay during the glacial and interglacial periods and found that sea ice reduced the effective wind area, thereby reducing the wave energy flux in the ice-free area. Liu et al. used the data of the seabed observation station under ice and ice-free conditions in a certain bay to systematically analyze and compare waves, residual water level, and residual current. The results showed that the coverage of ice has a significant inhibitory effect on the effective wave height and residual water level. In addition, the data from the seabed observation station showed that the sea ice coverage inhibited the resuspension of seabed sediments. Kagan et al. proposed that sea ice will not cause fundamental changes in tides.

[0013] Through the above analysis, the problems and defects of the existing technology are as follows: the existing technology has low monitoring accuracy for the distribution characteristics of sea ice in coastal waters. Summary of the invention

[0014] In order to overcome the problems existing in the related art, the disclosed embodiments of the present invention provide a method and system for evaluating sea ice condition influencing factors based on remote sensing.

[0015] The technical solution is as follows: A method for evaluating sea ice conditions influencing factors based on remote sensing, including: S1, select multi-year winter sea MODIS L1B data, extract multi-year winter sea ice area time series, use multi-year winter Sentinel-1A / B data based on gray level co-occurrence matrix calculation to extract sea ice texture features and identify sea ice types, and then supplement optical observation sea ice images; S2, using MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, the sea ice thickness is inverted to obtain the multi-year winter sea ice thickness and spatial distribution map, and the spatial distribution law of sea ice is analyzed; S3, using CSFR reanalysis wind field data and ERA5 reanalysis sea surface temperature data, and the three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center, obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice condition level; S4, based on the sea ice thickness inversion data, the Pearson-Ⅲ distribution model is used to obtain the spatial distribution of extreme sea ice thickness under different return periods.

[0016] In step S1, multi-year winter sea MODIS L1B data are selected and multi-year winter sea ice area time series are extracted, including: In optical imaging sea ice monitoring, the surface directional reflectivity is converted into spectral albedo, and the narrow-band reflectivity is converted into wide-band reflectivity; Narrow band reflectivity Defined as the upward radiation flux from a surface and downward radiated energy Ratio of: ; ; In the formula, is the multi-band reflectivity, is the multi-band upward radiation flux, is the downward radiation flux of the band, They are the minimum and maximum wavelengths of a certain band, respectively, the visible light band The range is 0.3μm-0.7μm; is the solar zenith angle, Band range ; use The three reflectivity values ​​are estimated and the obtained parameters are used to calculate the irradiance. Under the condition that the surface reflection conforms to Lambert's reflection law, the upward irradiance of the earth's surface and the downward irradiance of the earth's surface are: ; ; In the formula, is the irradiance without surface reflection, is the surface reflectivity, is the atmospheric hemispherical albedo, is the solar zenith angle, is the apparent downward irradiance, To observe the zenith angle, is the downlink irradiance, The path of the sun, is the atmospheric transmittance of the observation path; According to different reflectivity values, we can get Input the value of the variable, get the surface downlink irradiance and surface reflected irradiance, get the narrow-band albedo information of MODIS data, and use the narrow-band albedo and wide-band conversion formula of MODIS data: ; In the formula, It is the wide band of MODIS data. They are the narrow-band albedo corresponding to MODIS bands 1-5 and 7; For the inversion of sea ice thickness, Grenfell and Allison obtained a conversion model between wide-band albedo and sea ice thickness based on optical remote sensing theory. The formula is: ; In the formula, is the reflectivity of the spectrum of solar shortwave radiation, is the sea ice reflectivity corresponding to infinite ice thickness, is the correlation coefficient, , is the seawater albedo, is the thickness of sea ice, is the correlation coefficient of reflectivity, is the natural exponential function; Take 1.209, Taking 0.7, the root mean square error between the inverted sea ice thickness and the measured thickness is 0.02.

[0017] In step S1, multi-year winter Sentinel-1A / B data are used to extract sea ice texture features and identify sea ice types based on gray-level co-occurrence matrix operations, including: S101, using GLCM to obtain eight texture feature parameters of Sentinel-1 A / B images, including mean grayscale, variance, contrast, uniformity, dissimilarity, entropy, angular second moment, and correlation; the selected GLCM parameters include window size, direction angle, displacement, and grayscale quantization level; S102, sea ice sample collection, using the spectral characteristics of sea ice to identify ice skin, gray ice and white ice in winter sea areas, extracting sea ice types as samples from the ice skin, gray ice, white ice and water reflectance in the VV polarization mode of the ocean in Sentinel-1 A / B images; S103, classify sea ice based on a combination of multiple texture features, select the optimal sea ice feature texture combination and the optimal classification method based on the gray level co-occurrence matrix GLCM method suitable for extracting SAR image texture parameters, use four easy-to-identify samples to classify and identify sea ice, and count the sea ice area. The classification methods include support vector machine classification SVM, neural network classification method, and maximum likelihood supervised classification MLC.

[0018] In step S2, the sea ice thickness is inverted using MODIS L1B data based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model to obtain the multi-year winter sea ice thickness and spatial distribution map, and analyze the sea ice spatial distribution law, including: The winter sea ice conditions are divided into: light ice year, normal ice year obviously lighter, normal ice year slightly lighter, normal ice year slightly lighter, normal ice year, normal ice year slightly heavier, normal ice year heavier, normal ice year obviously heavier and heavy ice year; We analyzed the ice conditions in the years with heavy ice in the sea area, the ice conditions in different winter years, and the temporal and spatial distribution characteristics of sea ice. We used MODIS L1B optical remote sensing satellite data and Sentinel-1A / B SAR image data as data sources to classify and identify sea ice in the sea area in multi-year winters, analyze the development and change trends of multi-year sea ice, and obtain the maximum thickness of sea ice along the coast through the spatial distribution of the maximum value of multi-year sea ice thickness. Analyze the trend of multi-year sea ice development and change, including selecting statistics on the frequency and area of ​​sea ice in multi-year winters. The definition of the frequency of sea ice is: ; In the formula, is the probability of a pixel being frozen in the winter of a certain year, is the number of days in the remote sensing image where the pixel has ice formation in this winter. is the total number of this pixel in the remote sensing image in this winter.

[0019] In step S3, obtaining the relationship between air temperature and sea ice and ice condition level includes: Two parameters are introduced to measure the contribution of temperature to the development and melting of sea ice and its duration: cumulative frozen degree days CFDD and cumulative melt degree days CTDD, which represent the cumulative value below or above a critical temperature value over a period of time. The calculation formula is as follows: ; ; In the formula, For time, For the end time, is the start time, is the critical temperature for sea ice development and melting. ; It is the average daily temperature of the sea area on that day.

[0020] In step S3, obtaining the relationship between the atmospheric wind field and the sea ice and ice condition level includes: The CFSR wind field data is used to calculate the wind field distribution in the sea area in winter for many years, including the calculation of the total average wind field distribution in December, January, February and winter for many years.

[0021] In step S3, obtaining the relationship between salinity and sea ice and ice condition level includes: The salinity data in the sea area are selected for processing to obtain the average salinity of the sea area, including the calculation of the daily average salinity in December, January and February for many years.

[0022] In step S3, it also includes obtaining the relationship between the sea surface suspended matter concentration SSC and the sea ice and ice condition level, processing the MODIS L1B data into L2 level remote sensing reflection Rrs data, and performing geometric correction and atmospheric correction preprocessing on the data; the SSC inversion formula is: ; In the formula, is the concentration of suspended matter in seawater, is the natural index, is the reflectivity of the Rrs555 band; When the reflectivity meets the following four conditions, it is judged as sea ice: ; ; ; ; In the formula, is the normalized off-water reflectance, is the normalized off-water reflectance at 859 nm. is the normalized off-water reflectance at 555 nm. is the normalized off-water reflectance in the 412nm band, is the remote sensing reflectivity.

[0023] In step S4, based on the inversion data of sea ice thickness, the Pearson-III distribution model is used to obtain the spatial distribution of extreme sea ice thickness under different return periods, including: S401, inversion result processing, including removing outliers, resampling, and converting the inversion and resampling data of each year for many years into ASCII format data corresponding to coordinates and sea ice thickness values; S402, probability distribution theory and parameter estimation, choose Pearson-Ⅲ type to estimate the sea ice thickness return period; Pearson-Ⅲ type distribution is also called Γ type distribution, and the probability density function and distribution function are as follows: ; ; In the formula, is the frequency density function, for of Distribution function, is a random variable, is a positional parameter, is the shape parameter, is the scale parameter, is the probability distribution function, It is a multi-year maximum value. is the natural index; Assume that the annual maximum value of the sea ice thickness index is The probability density function of , its distribution function for: ; In the formula, is the probability distribution function; ; Annual maximum value The probability of occurrence is the guaranteed rate , and its calculation formula is: In the formula, is the probability guarantee rate, .

[0024] Another object of the present invention is to provide a remote sensing-based sea ice condition impact factor assessment system, which is used to assess the remote sensing-based sea ice condition impact factor assessment method, and the system comprises: The sea ice type recognition module is used to select multi-year winter sea MODIS L1B data and extract the multi-year winter sea ice area time series. It uses multi-year winter Sentinel-1A / B data to extract sea ice texture features and identify sea ice types based on gray-level co-occurrence matrix operations, thereby supplementing optical observation sea ice images. The module for obtaining the sea ice thickness and spatial distribution map of a certain sea in multi-year winter is used to use MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, to invert the sea ice thickness, obtain the multi-year winter sea ice thickness and spatial distribution map, and analyze the spatial distribution law of sea ice; The module for obtaining the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level is used to obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level by using CSFR reanalysis wind field data, ERA5 reanalysis sea surface temperature data and three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center. The module for obtaining the spatial distribution of extreme sea ice thickness in a certain sea under different return periods is used to obtain the spatial distribution of extreme sea ice thickness under different return periods based on the sea ice thickness inversion data and using the Pearson-Ⅲ distribution model.

[0025] Combined with all the above technical solutions, the beneficial effects of the present invention are as follows: the present invention selects MODIS L1B data of a sea area in the winter of 2003 to 2022, extracts the time series of sea ice area in a sea in the winter of nearly 20 years, and uses Sentinel-1A / B data from 2015 to 2022 to extract sea ice texture features and identify sea ice types based on gray level co-occurrence matrix (GLCM) calculation to supplement the optical observation sea ice image. The MODIS L1B data is used to invert the sea ice thickness of a sea based on the narrow band and wide band mutual conversion formula and the wide band albedo and sea ice thickness relationship model, and the sea ice thickness and spatial distribution map of a sea in the winter of 2003 to 2022 is obtained, and the spatial distribution law of sea ice is analyzed. Secondly, the CSFR reanalysis wind field data, the ERA5 reanalysis sea surface temperature data, and the three-dimensional temperature and salinity inversion field data provided by the Ocean Science Data Center are used to explore the relationship between air temperature, sea surface temperature, wind field and salinity and sea ice and ice conditions. Finally, based on the inversion data of sea ice thickness, the Pearson-Ⅲ distribution model is used to obtain the spatial distribution of extreme sea ice thickness in a certain sea under different return periods. And the monitoring accuracy is high.

[0026] The present invention can quickly form the range, thickness and change data of winter sea ice in a certain sea area, provide support and guarantee for navigation, breeding ranches, engineering facility safety and underwater operation related units, and avoid catastrophic accidents. For a long time, there has been insufficient research on the formation law of sea ice in mid-latitude waters at home and abroad, and the ability to predict its occurrence is relatively weak, which poses a great threat to marine engineering safety. The present invention focuses on the rapid identification and thickness calculation of winter sea ice, providing a new way to solve this problem.

[0027] Ice thickness parameter is a key indicator in offshore engineering design. Unreasonable indicators directly affect engineering safety and engineering cost, and have not been solved. This invention predicts the multi-year sea ice design parameters by inverting the sea ice area and thickness in a certain sea area in 20 years of winter, providing reliable parameters for engineering design and construction.

[0028] Different from previous studies, this invention combines multi-source remote sensing data and measured data and conducts comprehensive analysis to establish relevant models, obtains reliable time series data sets of sea ice area and thickness, and calculates high-precision sea ice monitoring data and multi-year engineering design parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description, serve to explain the principles of the present disclosure; Figure 1 This is a flow chart of a method for evaluating sea ice conditions influencing factors based on remote sensing provided by an embodiment of the present invention; Figure 2 are eight types of sea ice texture feature maps provided by an embodiment of the present invention; wherein a is an average grayscale feature map, b is a variance feature map, c is a uniformity feature map, d is a contrast feature map, e is a dissimilarity feature map, f is an information entropy feature map, g is an angular second-order moment feature map, and h is a correlation feature map; Figure 3 It is a sea ice level map of a certain sea winter from 2003 to 2022 provided by an embodiment of the present invention; Figure 4 It is a time series diagram of sea ice area and temperature of a certain sea from 2009 to 2010 provided by an embodiment of the present invention; Figure 5 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2009 to 2010; Figure 6 This is a time series diagram of sea ice area and temperature in a certain sea from 2010 to 2011; Figure 7 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2010 to 2011; Figure 8 This is a time series diagram of sea ice area and temperature in a certain sea from 2011 to 2012; Fig. 9 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2011 to 2012; Fig.10 The time series diagram of sea ice area and temperature in a certain sea from 2012 to 2013 is shown; Fig.11 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2012 to 2013; Fig.12 This is a time series diagram of sea ice area and temperature in a certain sea from 2015 to 2016; Fig.13 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2015 to 2016; Fig.14 This is a curve chart of the standard index of the maximum sea ice area in winter from 2003 to 2022, the standard index of the maximum daily average thickness in winter and the standard index of the average temperature in winter in a certain sea; Fig.15 This is a time series diagram of sea ice area and sea surface temperature in a certain sea during the winter of 2009-2010; Fig.16 This is a time series diagram of the average daily thickness of sea ice and sea surface temperature in a certain sea during the winter of 2009-2010; Fig.17 This is the change diagram of sea ice area, CFDD and CTDD in the winter of 2009-2010; Fig.18This is the change diagram of average sea ice thickness and CFDD and CTDD in winter 2009-2010; Fig.19 This is a diagram showing the fitting equation of sea ice area and accumulated temperature in a certain sea during the winter of 2009-2010; a is the diagram showing the fitting equation of sea ice area and CFDD during the freezing period, and b is the diagram showing the fitting equation of sea ice area and CFDD during the melting period; Fig. 20 This is a diagram showing the fitting equation of sea ice thickness and accumulated temperature in a certain sea during the winter of 2009-2010; a is the diagram showing the fitting equation of sea ice thickness and CFDD during the freezing period, and b is the diagram showing the fitting equation of sea ice thickness and CFDD during the melting period; Fig.21 This is the daily average salinity map of a certain sea area from 2016 to 2021; Fig. 22 It is the sea ice distribution map of a certain sea with a return period of many years; among them, a is the ice thickness distribution map with a return period of 1 year, b is the ice thickness distribution map with a return period of 5 years, c is the ice thickness distribution map with a return period of 10 years, d is the ice thickness distribution map with a return period of 25 years, and e is the ice thickness distribution map with a return period of 50 years. DETAILED DESCRIPTION

[0030] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.

[0031] The innovation of the present invention lies in that: the present invention selects the sea ice area time series of MODIS L1B in a certain sea area in winter for nearly 20 years, uses winter Sentinel-1A / B data to extract sea ice texture features and identify sea ice types based on gray level co-occurrence matrix (GLCM) calculation, and then supplements the optical observation sea ice data; uses MODIS L1B data to invert the sea ice thickness of a certain sea based on the narrow band and wide band mutual conversion formula and the wide band albedo and sea ice thickness relationship model, and obtains a 20-year winter sea ice thickness and spatial continuous change data set for the first time, and finally, based on the sea ice thickness inversion data, uses the Pearson-Ⅲ distribution model to obtain the spatial distribution of the extreme thickness of sea ice in a certain sea under different return periods.

[0032] Embodiment 1: The method for evaluating sea ice conditions influencing factors based on remote sensing provided by the embodiment of the present invention includes: S1, select multi-year winter sea MODIS L1B data, extract multi-year winter sea ice area time series, use multi-year winter Sentinel-1A / B data based on gray level co-occurrence matrix calculation to extract sea ice texture features and identify sea ice types, and then supplement optical observation sea ice images; S2, using MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, the sea ice thickness is inverted to obtain the multi-year winter sea ice thickness and spatial distribution map, and the spatial distribution law of sea ice is analyzed; S3, using CSFR reanalysis wind field data and ERA5 reanalysis sea surface temperature data, and the three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center, obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice condition level; S4, based on the sea ice thickness inversion data, the Pearson-Ⅲ distribution model is used to obtain the spatial distribution of extreme sea ice thickness under different return periods.

[0033] The present invention also provides a remote sensing-based sea ice condition impact factor assessment system, the system comprising: The sea ice type recognition module is used to select multi-year winter sea MODIS L1B data and extract the multi-year winter sea ice area time series. It uses multi-year winter Sentinel-1A / B data to extract sea ice texture features and identify sea ice types based on gray-level co-occurrence matrix operations, thereby supplementing optical observation sea ice images. The module for obtaining the sea ice thickness and spatial distribution map of a certain sea in multi-year winter is used to use MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, to invert the sea ice thickness, obtain the multi-year winter sea ice thickness and spatial distribution map, and analyze the spatial distribution law of sea ice; The module for obtaining the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level is used to obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level by using CSFR reanalysis wind field data, ERA5 reanalysis sea surface temperature data and three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center. The module for obtaining the spatial distribution of extreme sea ice thickness in a certain sea under different return periods is used to obtain the spatial distribution of extreme sea ice thickness under different return periods based on the sea ice thickness inversion data and using the Pearson-Ⅲ distribution model.

[0034] Embodiment 2, as another preferred embodiment of the present invention, the remote sensing-based sea ice condition impact factor assessment method provided in the embodiment of the present invention includes: 1. Materials and methods 1.1 Measured sea ice data.

[0035] Normally, the ice period in a certain sea area lasts for 2 to 3 months in winter each year. The sea ice observation data collected in a certain bay and surrounding waters are shown in Table 1. The measured data comes from a certain oil field marine environment service center. In the winter of 2006 to 2021, the first ice day in a certain bay is usually in early to mid-December. The specific first ice day varies from year to year. The earliest first ice day in the observation data is December 6, 2008, and the latest is January 6, 2016. Generally, the sea ice finally melts in mid-to-late February of the following year. The maximum observed sea ice thickness is 0.6m, and the accumulation height can reach 2m.

[0036] Table 1 Sea ice observation data collected in a bay and surrounding waters

[0037] 1.2 MODIS remote sensing image data.

[0038] The optical remote sensing data used in the present invention are MODIS-Aqua and MODIS-Terra data obtained from the Earthdata website of the National Aeronautics and Space Administration of a certain country (URL: https: / / ladsweb.modaps.eosdis.nasa.gov). MODIS data has the characteristics of short replay period, high spatial resolution and wide image range, and the spectral range of MODIS data is 0.4μm~12.4μm, covering the entire range from visible light band to thermal infrared band. Therefore, it is suitable to use MODIS remote sensing satellite data to monitor sea ice that changes rapidly and has a large range. Table 2 shows the characteristics of each band of MODIS remote sensing satellite. The present invention performs inversion statistics on the thickness of sea ice in a certain sea based on a multi-band combination.

[0039] Table 2 Introduction of MODIS bands

[0040] Table 2 (continued)

[0041] 1.2.1 MODIS optical remote sensing data preprocessing; MODIS data are divided into 6 levels. The MOD 021KM data used in the present invention belongs to the MODIS L1B level data product and has been radiometrically corrected. Therefore, only geometric correction and atmospheric correction need to be performed on the MODIS L1B data.

[0042] 1.2.1.1 Geometric correction: MODIS L1B data has been systematically geometrically corrected, and only geometric precision correction is required. By constructing a coordinate exchange relationship between the distorted image space and the correction space, the original image elements are transferred to the correction space, thereby achieving geometric precision correction.

[0043] 1.2.1.2 Atmospheric correction: When remote sensing satellites collect earth information, the scattering, reflection and adsorption phenomena in the atmosphere will affect the signal, so it is necessary to perform atmospheric correction on the image to obtain the original reflectivity of the ground object. The present invention uses the FLAASH (Fast line-of-sight atmospheric analysis of spectral hypercubes) atmospheric correction module to perform atmospheric correction on MODIS data.

[0044] 1.2.2 Inversion of sea ice thickness.

[0045] Generally speaking, the thicker the sea ice is, the stronger its reflection of sunlight is. Therefore, the thicker the sea ice is, the higher its reflectivity is. Grenfell et al. first concluded from the optical research on sea ice that the relationship between albedo and sea ice thickness is linearly increasing. Generally speaking, the albedo of sea ice is in the range of 0.11 to 0.7, but the albedo of snow is in the range of 0.8 to 1.0. When sea ice is covered with snow, the albedo increases. Therefore, it is necessary to extract the snow and select remote sensing satellite images with clear weather and no clouds over the sea as much as possible.

[0046] Remote sensing is the only practical means to map the albedo of land surfaces on a global scale, but the surface albedo is the main radiation uncertainty in current climate simulations. Therefore, in optical image sea ice monitoring, it is necessary to convert the surface directional reflectivity into spectral albedo, and narrowband into broadband albedo.

[0047] Narrow band reflectivity Defined as the upward radiation flux from a surface and downward radiated energy Ratio of: ; ; In the formula, is the multi-band reflectivity, are the minimum and maximum wavelengths of a certain band, respectively. is the multi-band upward radiation flux, is the solar zenith angle, Band range ; Use three reflectance values The three reflectivity values ​​are estimated and the obtained parameters are used to calculate the irradiance. Under the condition that the surface reflection conforms to Lambert's reflection law, the upward irradiance of the earth's surface and the downward irradiance of the earth's surface are: ; ; In the formula, is the irradiance without surface reflection, is the surface reflectivity, is the atmospheric hemispherical albedo, is the solar zenith angle, is the apparent downward irradiance, To observe the zenith angle, is the downlink irradiance, The path of the sun, is the atmospheric transmittance of the observation path; According to different reflectivity values, we can get Input the value of the variable, get the surface downlink irradiance and surface reflected irradiance, get the narrow-band albedo information of MODIS data, and use the narrow-band albedo and wide-band conversion formula of MODIS data: ; In the formula, It is the wide band of MODIS data. They are the narrow-band albedo corresponding to MODIS bands 1-5 and 7; For the inversion of sea ice thickness, Grenfell and Allison obtained a conversion model between wide-band albedo and sea ice thickness based on optical remote sensing theory. The formula is: ; In the formula, is the reflectivity of the spectrum of solar shortwave radiation, is the sea ice reflectivity corresponding to infinite ice thickness, is the correlation coefficient, , is the seawater albedo, is the thickness of sea ice, is the correlation coefficient of reflectivity, is the natural exponential function; Take 1.209, Taking 0.7, the root mean square error between the inverted sea ice thickness and the measured thickness is 0.02.

[0048] 1.3 Sentinel-1A / B remote sensing image data.

[0049] The selected SAR remote sensing satellite data is Sentinel-1 A / B IWS (Interferometric Wide Swath) Level 1 GRD data. The specific basic parameters are shown in Table 3. The selected data time series is from 2015 to 2022, including the complete sea ice period of a certain sea every year. Sentinel-1 consists of two polar-orbiting satellites A and B, with a replay period of 12 days. The spatial resolution of Sentinel-1 GRD data is 10m. Both satellite sensors are equipped with C band, and the polarization modes are VV and VH. The cross-polarization method can improve the distinction between seawater and sea ice, but the distinction between sea ice with different surface roughness is poor. Therefore, the accuracy of sea ice interpretation can be improved by comparing with optical images.

[0050] Table 3 Sentinel-1A / B image parameters

[0051] Table 3 (continued)

[0052] 1.3.1 Sentinel-1A / B image preprocessing

[0053] The present invention uses the application software SNAP (Sentinels Application Platform) developed by the European Space Agency to perform operations such as thermal noise removal, radiation correction, filtering and multi-view processing, geocoding and clipping on Sentinel-1 A / B data to obtain a grayscale image of the backscatter coefficient of a certain sea.

[0054] 1.3.2 Sentinel-1 image texture feature extraction.

[0055] In Sentinel-1 A / B images, the seawater has smooth and consistent texture features, while the sea ice surface has rough and disordered features. Therefore, sea ice recognition based on SAR satellite images usually relies on the study of texture features. Texture features are a way of expressing global features. The texture features of SAR images can reflect the spatial repeatability of image grayscale.

[0056] Gray Level Co-occurrence Matrix (GLCM) is a method suitable for extracting texture parameters of SAR images. Its principle is to use the correlation of gray space to explain texture features. The present invention uses GLCM to obtain the following eight texture feature parameters of Sentinel-1 A / B images.

[0057] (1) Average grayscale: It represents the average grayscale value of pixels in a fixed-size window, which can be used to understand the average brightness level in the window.

[0058] (2) Variance. It indicates the change of the gray value of pixels in the window. The larger the variance, the greater the fluctuation of the gray value of each pixel in the area.

[0059] (3) Contrast. The contrast is defined in the interval [0, (M-1) (N-1)], where M and N correspond to the number of rows and columns of the preset window, respectively. When the grayscale value of a pixel is at the upper or lower limit of the interval, it indicates that it has rich grayscale information, that is, high contrast. In order to ensure that the sea ice type in the sample is single and easy to identify, the contrast needs to be adjusted to meet the needs and improve the classification accuracy.

[0060] (4) Uniformity. This parameter reveals the texture similarity of the image and can evaluate the local changes in the texture. The value of uniformity is between [0,1]. The smaller the value, the less likely it is that the pixels in the image have the same grayscale value.

[0061] (5) Non-similarity: This parameter describes the irregularity of the grayscale value within the preset window.

[0062] (6) Entropy. Entropy is a measure of the amount of information in the selected area. The larger the entropy value, the more chaotic the texture distribution.

[0063] (7) Angular second-order moment: The angular second-order moment indicates the texture coarseness, fineness and distribution uniformity of the pixel grayscale value in the window.

[0064] (8) Correlation: The finer the texture, that is, the closer the pixel information, the higher the correlation.

[0065] When extracting features, it is necessary to select and adjust the parameters in the GLCM calculation process to ensure that the obtained features can accurately reflect the texture characteristics of different types of sea ice. The GLCM parameters selected by the present invention are as follows: (1) Window size. The window size should be selected according to the roughness of the target object. The rougher the texture, the larger the window should be. In the current research, the window size needs to be determined based on the scope of the study area and the image resolution. The window size selected in this invention is 7×7.

[0066] (2) Direction angle. Sentinel-1 A / B images are less affected by direction. The direction angle is usually expressed in an average or single manner. The direction angle selected in the present invention is 45°.

[0067] (3) Displacement. The size of the displacement mainly depends on the roughness of the image texture. When the image texture features are fine, a smaller displacement is selected, and when the texture features are relatively rough, a larger displacement is used. Based on previous research, the present invention selects a displacement of 2.

[0068] (4) Grayscale quantization level. The greater the quantization level, the higher the resolution of the grayscale image. However, too high a quantization level may result in too long a processing time. Therefore, the grayscale quantization level selected in the present invention is 64.

[0069] According to the GLCM operation parameters selected above, the eight sea ice texture features are obtained as follows: Figure 2 shown.

[0070] 1.3.3 Collection of sea ice samples.

[0071] In view of the limitation of spatial resolution of Sentinel-1 SAR images, the present invention utilizes the spectral characteristics of various types of sea ice to identify ice skin, gray ice and white ice in a certain sea area in winter. The sea ice types are extracted as samples from the ice skin, gray ice, white ice and water reflectance coefficient of the ocean in Sentinel-1 A / B images. However, the VH polarization mode of Sentinel-1 B images has poor information content and low image quality. Therefore, the water reflectance coefficient in the VV polarization mode is used to classify sea ice to improve its recognition and classification accuracy. The sea ice samples are shown in Table 4.

[0072] Table 4 Sea ice samples

[0073] 1.3.4 Classify sea ice based on a combination of multiple texture features.

[0074] Different types of sea ice have different surface texture features. The mean has a large difference in performance between seawater-ice skin-gray ice-white ice, which means that the mean has a greater effect on this classification. Dissimilarity performs well in the classification of white ice and gray ice, and uniformity can effectively distinguish between ice skin-gray ice-white ice. Therefore, the method of combining the three texture features of mean, dissimilarity and uniformity in the present invention selects the classification method with the highest accuracy to identify and classify sea ice in Sentinel-1 A / B images.

[0075] Support vector machine classification (SVM) is a method based on supervised classification, which is used to implement the classification task of two types of data. The key is to find the hyperplane with the longest interval among all the data to be classified, which enables the classification problem to be converted into a process of finding a convex quadratic programming problem.

[0076] The neural network classification method is a computer simulation method that simulates the human brain. It classifies images by simulating the recognition, memory and thinking process of neurons. In the classification that supports neural networks, the Logistic activation function is selected to identify and classify sea ice.

[0077] Maximum likelihood supervised classification (MLC) is a supervised classification method that classifies samples by evaluating the likelihood of each category and assigning them to the category with the highest likelihood.

[0078] According to the data characteristics of synthetic aperture radar images, semi-supervised classification methods or active learning classification methods are introduced into the classification research of synthetic aperture radar. The present invention selects the optimal sea ice feature texture combination based on the gray level co-occurrence matrix and selects the optimal classification method to classify and identify sea ice using a small number of identified samples, and then counts the sea ice area.

[0079] Vector machine classification will show that in SAR remote sensing images, the gray value of sea ice is larger and the gray value of sea water is smaller. At present, there is no detailed distribution and ice condition information of relevant sea ice, so optical remote sensing images are used to evaluate the accuracy of SAR image classification results. The maximum likelihood classification method based on texture features has higher classification accuracy for ice skin and white ice, but the recognition effect of gray ice is poor compared with the support neural network classification results; the support vector machine classification method misclassifies gray ice as ice skin, which reduces the sea ice classification accuracy; the recognition accuracy and sample quality of the neural network classification method are closely related to the number of iterations, and the recognition effect of gray ice is poor in the experiment. Among the above three methods, the maximum likelihood supervised classification has the best recognition effect for white ice. The main reason is that the gray value of white ice in SAR images is larger than that of other sea ice types, and it is easier to identify. Some data are disturbed by bad weather and speckle noise, which leads to an increase in the backscattering coefficient and a decrease in classification accuracy.

[0080] 1.4 ERA5 reanalysis data

[0081] The temporal resolution of ERA5 reanalysis data is 1 hour and the spatial resolution is 0.25°. ERA5 contains more than 200 parameters including air temperature, sea surface temperature, air pressure, etc., providing hourly atmospheric, land and ocean climate variables.

[0082] Based on the ERA5 reanalysis data, the present invention averages the 2m air temperature and the sea surface temperature from December to March of the following year in a certain sea study area from 2003 to 2023 to obtain the daily average temperature and the daily average SST.

[0083] 1.5 CFSR wind field data.

[0084] The sea surface wind field data uses the global atmospheric reanalysis product CFSR data provided by the National Center for Environmental Forecasting of a certain country. The sea surface wind field data provided by CFSR include the longitudinal and latitudinal wind speeds at a distance of 10m from the sea surface, with a spatial resolution of 0.205° and a temporal resolution of 6 hours.

[0085] 2. Analysis of sea ice conditions in a certain sea in winter.

[0086] The sea ice conditions of a certain sea are divided into: light ice year (1.0), obviously lighter than normal ice year (1.5), light ice year (2.0), slightly lighter than normal ice year (2.5), perennial ice year (3.0), slightly heavier than normal ice year (3.5), heavier than normal ice year (4.0), obviously heavier than normal ice year (4.5) and heavy ice year (5.0).

[0087] 2.1 Analysis of ice conditions in a certain sea area in an ice-heavy year.

[0088] like Figure 3 The winter sea ice level of a certain sea from 2003 to 2022 is shown. The year sequence of the ice level of a certain sea from 2003 to 2022 is plotted. It can be found that the ice level of the sea area shows obvious interannual changes, especially after 2013, the ice level shows a significant downward trend. Before 2013, the average ice level of a certain sea was 2.73. It is worth noting that in 2010, the ice level of a certain sea was 4.0, which was the year with heavy ice all year round, and it was the largest ice level in the past 20 years. After 2013, the average ice level of a certain sea was 2.05. In years with lighter ice conditions, due to the low resolution of optical remote sensing data, cloud cover and other data quality issues, as well as the short ice period and small ice area, it is impossible to use a single optical remote sensing image to obtain effective and continuous sea ice information. The present invention uses MODIS optical remote sensing data and Sentinel-1 A / B microwave remote sensing data to monitor sea ice in a certain sea.

[0089] Classification by sea ice level Figure 3 It can be seen that the sea ice condition level was above 3.0 for four consecutive years from 2010 to 2013, so the present invention analyzes the sea ice condition of a certain sea from 2010 to 2013. In addition, in order to verify the correctness of MODIS data, Sentinel-1 data was used to compare and analyze the sea ice area in winter 2016.

[0090] 2.1.1 Analysis of sea ice conditions in the winter of 2009-2010.

[0091] like Figure 4 The time series of sea ice area and temperature in a certain sea from 2009 to 2010 is shown in Figure 5 The following is a time series of the average daily thickness of sea ice and temperature in a certain sea from 2009 to 2010. The first ice formation date observed by remote sensing images was December 1, 2009, and the sea ice area was 913 km 2 The average daily thickness of sea ice was 0.015m, and then the temperature dropped slightly, and the sea ice area increased to 2757km on December 7. 2, the average daily thickness of sea ice reached 0.045m. After that, the temperature fluctuated, and the sharp drop in temperature was quickly reflected in the increase in sea ice area and the increase in average sea ice thickness. On January 6, 2010, the temperature dropped sharply to -13℃, and the sea ice area increased to 24504km 2 The average daily thickness of sea ice reached 0.143m. On January 15, the sea ice area in a certain sea was 24504km 2 The average daily thickness of sea ice was 0.143 m. On January 22, the sea ice area in a certain sea reached 28,347 km. 2 The average daily thickness of sea ice was 0.121m. On January 25, the sea ice in a certain sea reached its peak, with an area of ​​34,322km 2 , the average thickness of sea ice is 0.223m. It can be seen that the development of sea ice area is not synchronized with the increase of sea ice thickness. In the early stage of sea ice development, the increase of sea ice types is mainly ice skin (such as Figure 5 As shown in the figure, when the sea ice area is close to the peak, the sea ice thickness gradually increases, so the average daily thickness of the sea ice increases. After January 25, the temperature began to rise gradually. On January 27, the temperature was close to 0℃, reaching above freezing point. The sea ice area and the average daily thickness of the sea ice also decreased. The sea ice area on that day was 28601km 2 , the average daily thickness of sea ice was 0.141m. After that, the temperature showed an overall upward trend but still had twists and turns, and the sea ice area and the average daily thickness of sea ice also showed a small fluctuation phenomenon. After that, the temperature rose above 0℃ for a period of time. During this period, the sea ice began to melt faster, and the sea ice area and the average daily thickness of sea ice decreased rapidly. According to MODIS image observations, March 12 was the last ice day in a certain sea, and the sea ice area on that day was 2927km 2 The average daily thickness of sea ice is 0.021m.

[0092] The ice condition level of a certain sea in the winter of 2009-2010 was 4.0, which was the highest ice condition level in the sea ice monitoring in the past 20 years. According to MODIS image statistics, the first ice day of this year was December 1, and the last ice day was March 12, with a total ice period of 102 days. After the temperature reached the lowest in January, the sea ice area and the average daily thickness of sea ice increased rapidly. After that, the temperature rose slightly, but the sea ice area and sea ice thickness continued to increase. This shows that when the temperature is lower than the freezing temperature of seawater, the sea ice area and sea ice thickness will continue to increase.

[0093] According to the inversion results of sea ice thickness from 2009 to 2010, sea ice is mainly concentrated in the first bay, which is related to the higher latitude of the first bay and the lower temperature than the second and third bays; the second is the second bay, and the third is the smallest. Sea ice first formed in the first bay, mainly ice skin with a thickness of less than 0.05m. In the MODIS inversion results on December 19, sea ice appeared in the second and third bays, but the thickness was small, and the type of sea ice was mainly ice skin. In mid-December 2009, the temperature dropped sharply, and sea ice began to develop over a large area. On December 31, a large area of ​​white ice was formed in the northern and eastern coasts and the center of the first bay. At this time, the sea ice of the second and third bays was mostly developed along the coastline. On January 25, the sea ice area of ​​the sea reached its peak, almost occupying the entire third bay and the second bay of the second bay, and white ice developed along the southern coast of the second bay and the southern coast of the third bay. Subsequently, the sea ice area in Bay 2 and Bay 3 decreased, while sea ice developed in Bay 1. On February 12, the sea ice area in Bay 1 reached its peak, with gray ice and gray-white ice being the main types of sea ice. Large-scale white ice developed in the northeast of Bay 1. In the observation images on February 20, gray ice, gray-white ice and white ice decreased, and the average daily thickness of sea ice also decreased to 0.103m. Subsequently, the temperature gradually increased, and the sea ice began to melt. On March 7, sea ice existed only in the eastern part of Bay 1, mainly ice skin.

[0094] 2.1.2 Analysis of sea ice conditions in the winter of 2010-2011.

[0095] The sea ice area, daily average sea ice thickness and daily average temperature of a certain sea in 2010~2011 are as follows: Figure 6 The time series diagram of sea ice area and temperature in a certain sea from 2010 to 2011 is shown in Figure 7 The time series of the average daily thickness of sea ice and temperature in a certain sea from 2010 to 2011 is shown in Figure 2. The first ice day observed by remote sensing images was December 14, and the sea ice area had reached 3191 km 2 The average daily thickness of sea ice was 0.066m. The temperature dropped to -9℃ on December 15 and then rose again, but it still fluctuated around 0℃. The sea ice area grew slowly and reached 10690km on January 12. 2 The average daily thickness of sea ice increased to 0.101m. The temperature dropped to -12℃ on January 17, but the sea ice area reached its maximum value in the observation image on January 21. The temperature on January 21 was -7.5℃ and the sea ice area was 23858km 2 On that day, the average daily thickness of sea ice also reached a maximum of 0.209m. This shows that the impact of temperature rise and fall on the sea ice area and the average daily thickness of sea ice has a certain lag effect. After January 17, the temperature showed an overall upward trend, the sea ice began to melt, and the sea ice area and the average daily thickness of sea ice gradually decreased. On February 7, the sea ice area was reduced to 5657km 2The average daily thickness of sea ice dropped to 0.087m. From February 7 to February 14, the temperature dropped again, reaching -8.6℃ on the 14th. Although the temperature gradually rose afterwards, it was still near the freezing point, causing the sea ice to refrozen. On March 3, the sea ice area reached 8973km 2 The average daily thickness of sea ice is 0.475m. Due to cloud cover and low temporal resolution, the quality of MODIS images obtained in the winter of 2010-2011 was low, and no sea ice was observed in MODIS images after March 3.

[0096] According to MODIS remote sensing image statistics, the sea ice period in the winter of 2010-2011 was 80 days, which was shorter than the sea ice period in the winter of 2009-2010, but the average sea ice thickness was close to that in the winter of 2009-2010.

[0097] According to the results of sea ice thickness inversion from 2010 to 2011, sea ice began to develop along the northern coast of Bay 1, the southern coast of Bay 2, and the northern coast of Bay 3. As the temperature dropped, the sea ice area in Bay 1 expanded and the ice thickness increased in late December, mainly gray ice. The sea ice area in Bays 2 and 3 did not increase significantly, but the sea ice thickness increased. The sea ice thickness on the first ice day was below 0.1m, and it developed to be mainly gray ice with a small amount of white ice. On January 21, the sea ice in Bay 1 reached its peak, and the sea ice almost occupied the entire Bay 3. The sea ice in Bay 3 was mainly gray ice, and a large amount of sea ice was generated in the southern part of Bay 3. The sea ice in the south was more serious than that in the north. The north was mostly ice skin, while the south was mainly gray ice and gray-white ice. The sea ice area in Bay 1 was the largest, mainly gray ice. Subsequently, the sea ice gradually melted and the sea ice area decreased, but on February 12, the sea ice in Bay 1 intensified again, with the white ice area decreasing and the gray ice area increasing. On March 3, sea ice only existed in a certain bay, and no sea ice was observed in MODIS images after March 3. Overall, the trend of sea ice first increased and then decreased, but in early February, a certain bay showed a phenomenon of refrozen.

[0098] 2.1.3 Analysis of sea ice conditions in the winter of 2011-2012.

[0099] The series of sea ice area, daily average sea ice thickness and daily average temperature in a certain sea from 2011 to 2012 are as follows: Figure 8 This is a time series diagram of sea ice area and temperature in a certain sea from 2011 to 2012. Fig. 9 The following is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2011 to 2012. December 11 was the first ice day observed by MODIS remote sensing images, with a sea ice area of ​​834 km 2 The average daily thickness of sea ice was 0.067m. The temperature then dropped sharply to -6.79℃ on December 15, but the sea ice area did not increase significantly. On December 15, the sea ice area was 1115km2 The average daily thickness of sea ice was 0.059m. After that, the temperature fluctuated around -3℃, negative accumulated temperature was effectively accumulated, and the sea ice area began to increase gradually. On January 13, the sea ice area reached 13373km 2 The average daily thickness of sea ice reached a maximum of 0.201m in the winter of 2011-2012. On January 22, 2012, the temperature dropped sharply to -10.8℃, the first time the temperature dropped below -10℃ this winter. After the temperature dropped rapidly, the sea ice development speed accelerated and the sea ice area increased. On January 24, the MODIS image showed that the sea ice area in a certain sea was 23775km 2 The average daily thickness of sea ice reached 0.1486m. The temperature rose after that, but remained below -3℃ overall, causing the sea ice area to expand further. On February 2, the sea ice area reached the largest in the entire winter, reaching 24,770km 2 The average daily thickness of sea ice was 0.091m. The temperature on February 14 was -1.1℃, higher than the growth temperature of sea ice (-1.8℃), and the sea ice began to melt. On February 18, the sea ice area was 11853km 2 The average daily thickness of sea ice was 0.083m. During this period of gradual warming, the sea ice also showed a trend of gradual melting. On February 28, the sea ice area was reduced to 3340km 2 The average daily thickness of sea ice was 0.087 m. March 2 was the last day of sea ice observed by MODIS remote sensing data. The sea ice area on that day was 2008 km 2 The average daily thickness of sea ice is 0.073m.

[0100] The results of the inversion of sea ice thickness from 2011 to 2012 show that the first ice day of this winter is December 9. On that day, sea ice only developed in the northern part of Bay 1 and the thickness was small. It was not until January 5, 2012 that sea ice was observed in the southern part of Bay 2 and Bay 3 in remote sensing images, and the thickness of the sea ice was about 0.05m. The sea ice area reached its maximum on January 24, and the sea ice type was mostly gray-white ice of 0.15-0.3m. The sea ice in the eastern part of Bay 1 was thicker than that in the western part of Bay 1, and white ice developed near the coastline. Subsequently, the sea ice gradually melted, and the melting degree of the sea ice in Bay 1 was also greater in the western part than in the eastern part. On February 16, the sea ice refrozen, a large area of ​​white ice formed along the northern coast of Bay 1, and the sea ice in Bay 2 and Bay 3 also intensified to varying degrees. Afterwards, as the temperature rose, the island entered the melting period again, and the melting of sea ice accelerated. The last ice-ending day of the observed island was March 2, with only a small area of ​​gray ice and ice skin on the northern coast and eastern part of a certain bay.

[0101] 2.1.4 Analysis of sea ice conditions in the winter of 2012-2013.

[0102] The time series of sea ice area, daily average sea ice thickness and daily average temperature in a certain sea in the winter of 2012-2013 are as follows: Fig.10 The time series diagram of sea ice area and temperature in a certain sea from 2012 to 2013 is shown in the figure. Fig.11 This is a time series diagram of the average daily thickness of sea ice and temperature in a certain sea from 2012 to 2013.

[0103] The first ice date observed by MODIS images this winter was December 9, which was earlier than in 2011 and 2012 when the ice age was 1.0. At the beginning of this ice age, the temperature dropped to -7.9℃, and then fluctuated greatly for a period of time, but generally remained below -2℃. The long-term low temperature caused the sea ice area to expand rapidly. As of January 14, the sea ice area was 20449km 2 The average daily thickness of sea ice reached 0.216m. After that, the temperature continued to remain below -4℃, and the sea ice area accumulated. The sea ice reached its peak on January 18, with an area of ​​30,751 km 2 , the average daily thickness of sea ice was 0.229m, after which the temperature rose to -2℃, the sea ice began to melt, and the sea ice area began to decrease. On January 25, the sea ice area decreased to 19037km 2 The average daily thickness of sea ice also decreased to 0.085m. Then the temperature continued to rise and the sea ice area continued to decrease. By February 1, the sea ice area dropped to 10190km 2 The melting of sea ice is also reflected in the decrease in sea ice thickness. On February 1, the average daily thickness of sea ice was 0.073m. On February 7, the temperature dropped to -11.9℃. The sudden low temperature caused the sea ice to stop melting and then began to grow. The sea ice area in a certain sea increased. On February 13, the sea ice area was observed to increase again to 23051km 2 The average daily thickness of sea ice also increased to 0.128m. After that, the temperature showed an overall upward trend, and the sea ice melted faster. March 1 was the last ice day of the year observed by MODIS images.

[0104] The winter of 2012-2013 was a slightly heavy year for sea ice. The ice period observed in remote sensing images was 82 days, close to previous years, but the maximum sea ice area was larger. The trend of sea ice area change in this winter showed a pattern of first increasing, then decreasing, then increasing, and then decreasing. The reason is that the sharp drop in temperature in early February 2013 was conducive to the continued development of sea ice.

[0105] The inversion results of sea ice thickness in a certain sea in the winter of 2012-2013 show that the first ice day observed on MODIS images was December 9, 2012. On this day, there was sea ice in the three bays of a certain sea. The sea ice area in Bay 1 was larger than that in Bay 2 and Bay 3. The sea ice in Bay 2 and Bay 3 developed in the coastal area. Since Bay 3 and Bay 2 were blocked by clouds, only sea ice in Bay 1 was observed on December 18. Judging from the sea ice in Bay 1, the thickness of sea ice decreased. The temperature in a certain sea dropped sharply in late December, and the sea ice grew rapidly. On December 30, there was a large area of ​​white ice in Bay 1 with less ice skin. The southern part of Bay 2 and Bay 3 was also dominated by gray ice. On January 18, 2013, the sea ice area reached its peak. The sea ice in Bay 3 was the thinnest, mostly below 0.1m. The sea ice in Bay 2 and Bay 1 was almost all above 0.1m, and there was a lot of white ice on the northern coast of Bay 1. When the image of Bay 1 was observed on February 1, the sea ice area had already decreased. On February 13, the sea ice area in Bay 2 and Bay 3 decreased greatly. Although the white ice area in Bay 1 decreased, the sea ice area increased and refrozen still occurred. After that, the sea ice in Bay 2 and Bay 3 disappeared completely, and the sea ice in Bay 1 gradually melted. The distribution of sea ice in the two images on March 1 and March 3 was similar, both in the middle of Bay 1.

[0106] 2.1.5 Analysis of sea ice conditions in the winter of 2015-2016.

[0107] Since MODIS images are easily affected by clouds and fog and have a low temporal resolution, the present invention uses Sentinel-1 A / B images to extract the sea ice area of ​​a certain sea to supplement the winter sea ice age data. At the same time, the sea ice area extracted from Sentinel-1 A / B data is combined with the data in the "Marine Disaster Report" issued by the Ocean Administration to invert the sea ice thickness from MODIS data and verify the sea ice area obtained by statistics.

[0108] Due to the long replay period of Sentinel-1 A / B satellites and the small observation range, one Sentinel-1 image cannot cover the entire sea area, so the image closest to the observation date is selected for comparison. The image of Bay 1 is selected as the observation image on February 9, 2016, and the image of Bay 2 is selected as the observation image on February 7, 2016. Bay 3 was observed within 10 days, so the sea ice image of Bay 3 was not selected.

[0109] By comparing the spatial distribution of sea ice in a certain sea in the MODIS optical image on February 10, 2016 with the spatial distribution of sea ice in the Sentinel-1 SAR image around February 8, 2016, it can be seen that both MODIS and Sentinel-1 A / B can effectively monitor sea ice. As an optical remote sensing satellite, MODIS can provide richer sea ice information, such as color, shape and thickness, but as an optical image, it is greatly affected by weather and clouds; Sentinel-1 A / B data can monitor a wider ocean area and is not affected by bad weather. The imaging time is continuous. Compared with the inversion results of MODIS L1B data, in Sentinel-1 A / B images, sea water, ice skin, gray ice and white ice have obvious grayscale value differences, which can be effectively distinguished using the maximum likelihood classification method based on texture feature extraction. However, there were also cases of misclassification of thick gray ice and white ice. Apart from the difference in sea ice classification caused by the fact that Sentinel-1A / B images and MODIS L1B images were not imaged at the same time, the reason for the difference is that the winter environment of a certain sea is changeable. For example, wind and waves on the sea surface will increase the roughness of the sea surface, making the scattering of SAR waves on the sea surface more complex and diverse, resulting in an increase in the backscattering coefficient. Especially in the case of strong winds, the waves on the sea surface will become more violent, resulting in an increase in the backscattering coefficient of seawater in SAR images. When classifying sea ice, seawater will be misclassified as white ice.

[0110] The series of sea ice area, daily average sea ice thickness and daily average temperature in a certain sea from 2015 to 2016 are shown in the figure below: Fig.12 The time series of sea ice area and temperature in a certain sea from 2015 to 2016 is shown in Fig.13 The average daily thickness of sea ice and temperature in a certain sea from 2015 to 2016 are shown in the time series diagram. From the Sentinel-1 A / B remote sensing images, it is observed that the first ice day of this year is December 23, 2015, and the sea ice area is 3147km 2 The first ice date of MODIS is December 27, and the sea ice area is 4325km 2 The average daily thickness of sea ice on that day was 0.081m. From December 23 to December 26, the temperature dropped sharply from 0℃ to -8℃. The sharp drop in temperature is conducive to the formation of sea ice. The first ice day observed by MODIS images this winter is later than the first ice day observed by Sentinel-1 A / B images, and the sea ice area of ​​a certain sea on December 23 observed by Sentinel-1 A / B images is 3147.16km 2 MODIS images showed that the sea ice area in a certain sea was 4325 km on December 27. 2The temperature dropped to -11.9°C on January 23, the lowest temperature in an ice year with an ice condition level greater than 1.0 since the winter of 2009. The sea ice area observed on January 25 was 18,560 km 2 The average daily thickness of sea ice is 0.086m. On February 1, the maximum sea ice area of ​​a certain sea was observed to be 28099km. 2 The average daily thickness of sea ice was 0.092m, and the sea ice area was much smaller than that in previous years with normal ice and slightly heavier years with normal ice, which shows that temperature is not the absolute factor that determines the development and growth of sea ice. After that, the temperature rebounded as a whole, rising above 0℃ on February 10, and then fluctuated greatly. The temperature dropped to -7℃ on February 14 and then rose to 0℃ on February 19. The sea ice area had shrunk to 5279km on February 17. 2 , no sea ice was observed in MODIS images after February 29.

[0111] The sea ice area observed by MODIS on January 19 was 17956 km 2 On January 20, the sea ice area observed by Sentinel-1 images was 16990.6 km 2 , the sea ice area differs by 964km 2 , ignoring the melting of sea ice within two days, the error is consistent with the actual situation.

[0112] The ice period observed in remote sensing images in the winter of 2015-2016 was 64 days, which was relatively short, but the growth rate of sea ice area was relatively fast this winter. The average daily thickness of sea ice showed an overall trend of increasing first, then decreasing, then increasing, and then decreasing again. However, when the sea ice area reached its peak, the average daily thickness of sea ice on that day was not the maximum value of sea ice thickness this winter.

[0113] 2.2 Temporal and spatial distribution characteristics of sea ice in a certain sea.

[0114] Using MODIS L1B optical remote sensing satellite data and Sentinel-1A / B SAR image data as data sources, the sea ice in a certain sea area in the winter of 2003~2022 was classified and identified, and the development and change trend of the sea ice in a certain sea in the past 20 years was analyzed.

[0115] Through the spatial distribution of the maximum sea ice thickness in a certain sea from 2003 to 2022, it can be found that the maximum thickness of sea ice along the coast of a certain sea is larger, with a thickness value of 0.7~1.2m, the maximum thickness of sea ice in the center of a certain sea is small, with a thickness value of 0.1~0.4m, and the area of ​​sea ice with a maximum thickness of more than 1m in a certain bay is the largest, mostly concentrated in the northern part of a certain bay and coastal waters.

[0116] Sea ice in a certain sea usually reaches its peak from the end of January to the beginning of February, and then the area begins to decrease as the temperature rises. The thickness of sea ice gradually decreases from the coast to the central area. When the sea area of ​​a certain bay freezes, it moves from north to south and from nearshore to farshore, and when it melts, it moves from west to east and from farshore to nearshore. The thickness of sea ice in a certain bay is mainly concentrated between 0.15 and 0.3m; sea ice with a thickness of more than 0.3m mostly appears in January and February, and is usually distributed along the eastern coast, the northern coast, and where sea ice is dense. The thickness of sea ice in a certain bay is distributed as follows: the closer to the coast, the thicker it is, and the southern coast is thicker than the northern coast. The central part of the sea will freeze in years when the ice level is slightly heavier than the normal ice year, and usually does not freeze in other years. When it melts, it moves from north to south and from farshore to nearshore. The thickness of sea ice in a certain bay is usually below 0.15m, and the sea ice types are mainly ice skin and gray ice. During the peak ice period, white ice is formed on the southern coast of a certain bay. The sea ice development process in Bay 3 is similar to that in Bay 2.

[0117] Overall, the main ice formation occurred in the first bay of the Bay. The thickness and area of ​​sea ice in the Bay were the largest, followed by the Bay, and the smallest in the Bay. This follows the distribution law of sea ice in latitudes with severe ice formation at high latitudes. However, the sea ice area and thickness of the Bay were greater than those of the Bay in 2006, 2010, 2011, and 2013. Although the latitude of the Bay is lower than that of the Bay, the water depth of the Bay is less than that of the Bay. In addition, there are many freshwater rivers such as the Yellow River in the Bay, which makes the salinity of the sea water in the Bay relatively low. Therefore, sea ice is more likely to develop in the Bay. It can be seen that the average thickness of sea ice in January and February is greater than that in December. The closer the sea ice is to the shore, the greater the average ice thickness. On the contrary, the farther from the shore, the thinner the ice layer. There is usually no sea ice development in the center of the Bay. The average ice thickness of the Bay is the largest, followed by the Bay, and the average ice thickness of the Bay is the smallest. The average thickness of sea ice in a certain bay in the east in winter is greater than that in a certain bay in the west.

[0118] Because it is controlled by many factors, it is not accurate to measure the distribution, growth and melting of sea ice only by the sea ice area. In order to fully understand the sea ice situation, the frequency and area of ​​sea ice in a certain sea during the winter of 2003 to 2022 were selected for statistics. The definition of the frequency of sea ice is: ; In the formula, is the probability of a pixel being frozen in the winter of a certain year, is the number of days in the remote sensing image where the pixel has ice formation in this winter. is the total number of this pixel in the remote sensing image in this winter.

[0119] From the spatial distribution of the frequency of ice formation in a certain sea from 2003 to 2022 (%), it can be seen that the ice formation area in Bay 1 is the largest, followed by Bay 2, and the smallest is Bay 3. The annual ice formation frequency gradually increases from the coastal area to the center of a certain sea. Overall, the ice formation frequency in Bay 1 is the largest among the three bays. More than half of the sea area in Bay 1 has an ice formation frequency of more than 70%, so the main ice formation area in a certain sea is Bay 1. The main ice formation areas of Bays 2 and 3 are mainly in the southeast coastal waters, with ice formation frequencies between 60% and 80%, which is significantly lower than that of Bay 1. The ice formation frequency of a certain sea obtained by remote sensing images in winter from 2003 to 2022 is 0 in the middle of a certain sea and the strait of a certain sea. The ice formation range of a certain sea is the largest in February. The size distribution of the annual average ice formation frequency also reflects the development process of sea ice growth and melting in a certain sea every winter. Sea ice first develops in the coastal waters of the bay. As the temperature drops, the sea ice gradually develops toward the center of the sea. The sea ice in the southeast develops faster than the sea ice in the northwest.

[0120] The area and thickness of sea ice are mainly affected by environmental factors such as temperature, while the spatial distribution of area and thickness is controlled by other factors. When the temperature drops below the sea ice development temperature, the sea water loses a lot of heat and sea ice begins to form. The maximum water depth of a certain sea area is 90m and the average water depth is 18m. Compared with other ice-covered sea areas, the average water depth of a certain sea is smaller. Without considering factors such as temperature and seawater salinity, the spatial distribution of sea ice area and sea ice thickness is mainly affected by the heat capacity of seawater. The size of seawater heat capacity mainly affects the difficulty of seawater freezing, the vertical distribution of sea ice, and the plane distribution of sea ice. The heat capacity of a specific substance can be calculated by multiplying its mass and specific heat capacity. Assuming that the specific heat capacity of seawater at each depth in each area of ​​a certain sea area remains unchanged, the size of the specific heat capacity of seawater is determined by the quality of seawater, and the depth of seawater is the direct factor controlling the quality of seawater. From the spatial distribution of the frequency of ice formation in a certain sea (%) from 2003 to 2022 and the water depth (m) of a certain sea area, it can be seen that the frequency of ice formation in the sea area with a sea depth of less than 15 m in a certain sea area reached 70%. The smaller the water depth in the sea area, the higher the frequency of ice formation. Therefore, the spatial distribution of sea ice in the horizontal direction is closely related to the water depth. The water depth of the southwest coast of Bay 2 and Bay 3 is shallow, which also explains the phenomenon that the frequency of ice formation in the southwest coast of Bay 2 and Bay 3 is higher.

[0121] 3. Analysis of factors affecting sea ice in a certain sea.

[0122] The present invention refers to the day with the maximum sea ice area when sea ice is observed in MODIS optical images each year as the freezing period, and the period from the observation that the sea ice area decreases to the last observation of sea ice is called the melting period.

[0123] 3.1 Effect of temperature on sea ice conditions in a certain sea.

[0124] 3.1.1 Air temperature. The sea ice in a certain sea is sea ice in a shallow inland sea and is significantly affected by meteorological conditions. Changes in the sea ice conditions in a certain sea are closely related to cold air activities from December to March of the following year. The present invention uses ERA5 2m air temperature reanalysis data. The 2m air temperature data in a certain sea from the first ice day to the last ice day in each winter from 2003 to 2023 are selected for processing, and the average winter temperature in a certain sea is obtained. It is normalized with the maximum sea ice freezing area in a certain sea in winter from 2003 to 2023 and the maximum daily average thickness in a certain sea in winter from 2003 to 2023, and plotted into a line graph. By Fig.14 From the standard index curves of the maximum sea ice area in winter from 2003 to 2022, the standard index of the maximum daily average thickness in winter and the standard index of the average temperature in winter, it can be seen that the temperature has a significant negative correlation with the ice area and the maximum daily average thickness.

[0125] 3.1.2 Sea surface temperature. Generally speaking, when the sea surface temperature is below 0°C, the sea water will gradually freeze to form an ice layer, and when the sea temperature is above 0°C, the ice layer will gradually melt when it comes into contact with the sea water. In addition, changes in sea surface temperature will also affect the shape and distribution of sea ice. For example, in areas with higher sea temperatures, sea ice is more likely to form thin ice or melt, while in areas with lower sea temperatures, it is more likely to form thicker ice layers. The present invention uses ERA5 sea surface temperature data to explore the relationship between the sea surface temperature and sea ice conditions in a certain sea.

[0126] Taking 2009-2010 as an example, the relationship between the sea surface temperature and sea ice conditions in a certain sea is analyzed. Fig.15 This is a time series diagram of sea ice area and sea surface temperature in a certain sea during the winter of 2009-2010. Fig.16 The time series of the average daily thickness of sea ice and sea surface temperature in a certain sea during the winter of 2009-2010 shows that the sea surface temperature of a certain sea continued to decrease during the winter of 2009-2010. The sea surface temperature was 7.18℃ on the first ice day, and it had dropped to 0.48℃ on the peak ice day. The sea surface temperature then continued to drop, dropping to the lowest sea surface temperature of this winter -0.51℃ on February 14, but the sea ice area and average sea ice thickness had decreased. The sea surface temperature in the winter of 2009-2010 showed a trend of first steadily decreasing and then slowly increasing, with a small fluctuation and temperature difference, and the sea ice area, average daily sea ice thickness and sea surface temperature were negatively correlated overall, indicating that the sea surface temperature has an impact on sea ice but is not an absolute dominant factor.

[0127] 3.1.3 Accumulated temperature. A decrease in temperature usually leads to an increase in sea ice area, but not all cases will an increase in temperature lead to a decrease in sea ice area and thickness. Generally speaking, when the temperature is lower than the critical value for sea ice development and melting, the sea ice will begin to grow, resulting in an increase in sea ice area; when the temperature is higher than the critical temperature, the sea ice begins to melt and the sea ice area will decrease. The present invention introduces two parameters for measuring the contribution of temperature to the development and melting of sea ice and the duration of the temperature: cumulative frozen degree days (CFDD) and cumulative melt degree days (CTDD), which represent the cumulative values ​​below or above a certain critical temperature value over a period of time, and the calculation formula is as follows: ; ; In the formula, For time, For the end time, is the start time, is the critical temperature for sea ice development and melting. ; It is the average daily temperature of the sea area on that day.

[0128] 3.1.4 Correlation analysis of meteorological factors. Sea ice formation and melting is an interactive system, which is jointly affected by multiple factors in the environment (air temperature, sea surface temperature, CFDD, CTDD). Therefore, when exploring the formation and melting process of sea ice, it is necessary to fully consider the comprehensive impact of these factors on sea ice. The sea ice growth and melting process is divided into the freezing period and the melting period, and the correlation analysis is performed on the daily average air temperature, sea surface temperature SST, CFDD, and CTDD of a certain sea. The analysis results are shown in Table 5 below.

[0129] Table 5 Correlation coefficients between freezing and melting periods of a certain sea and temperature factors

[0130] Note: ** indicates significant correlation at the 0.01 level (two-tailed); * indicates significant correlation at the 0.05 level (two-tailed).

[0131] As shown in Table 5, when the daily average temperature and SST are used as meteorological factors for correlation analysis with the sea ice area in the ice period, freezing period and melting period, the correlation coefficients are all negative, indicating that the lower the temperature or SST, the larger the sea ice area. After the sea ice period is divided into freezing period and melting period, the meteorological factor with the greatest correlation with the sea ice area in the freezing period is CFDD (r=0.844), and the meteorological factor with the greatest correlation with the sea ice area in the melting period is CTDD (r=-0.524), indicating that the growth of the sea ice area in the entire sea is more correlated with the accumulation of melting below the critical value of seawater freezing. When the correlation analysis is conducted on the single-day temperature as a meteorological factor with the sea ice thickness, the correlation coefficients with the sea ice thickness in the total ice period, freezing period and melting period are all negative but the correlation is small. Among them, the correlation coefficient between SST and sea ice thickness in the melting period is the largest (r=-0.130), indicating that the correlation between the single-day temperature and the sea ice thickness is not large. However, during the freezing period, the correlation coefficient between CFDD and sea ice thickness is r=0.599, which is a significant positive correlation, that is, the greater the temperature accumulation value below the critical temperature of seawater freezing, the thicker the sea ice; during the melting period, the correlation coefficient between CTDD and sea ice thickness is the largest (r=-0.248) and is negatively correlated, that is, the greater the temperature accumulation value above the critical value of seawater freezing, the thinner the sea ice.

[0132] In summary, during the freezing period, CFDD is the meteorological factor with the highest correlation with sea ice area and sea ice thickness; on melting days, CTDD is the meteorological factor with the highest correlation with sea ice area and sea ice thickness. Fig.17 This is the change diagram of sea ice area, CFDD and CTDD in winter 2009~2010. Fig.18 This is the change diagram of average sea ice thickness and CFDD and CTDD in winter 2009-2010; Taking the winter of 2009-2010 as an example, the sea ice area showed a trend of increasing first, reaching the maximum on the ice peak day and then starting to decrease. During the freezing period, CFDD accumulated rapidly and the sea ice area also showed a trend of gradually increasing. The growth rate and trend of the two were almost the same. At the same time, CTDD showed an almost stagnant phenomenon. When the growth rate of CTDD accelerated, the sea ice area decreased rapidly. The daily average thickness of sea ice and the changes of CFDD and CTDD over time are shown in the figure above. The daily average thickness of sea ice showed an overall trend of increasing first and then decreasing. However, during the initial ice period to the ice peak period, CFDD continued to grow while the daily average thickness of sea ice decreased on January 22, from 0.1434 m on January 15 to 0.1208 m. The reason is that the expansion of sea ice area during this period mostly relied on the formation of thinner ice skin. With the increase of CFDD, the sea ice area increased while the thinner ice layer developed, and the overall thickness of sea ice increased. By January 25, the daily average thickness of sea ice increased to 0.2228 m. Subsequently, CTDD increased, sea ice began to melt rapidly, and the average thickness of sea ice decreased.

[0133] In order to further explore the correlation between CFDD, CTDD and sea ice conditions, the sea ice area and sea ice thickness during the freezing period and melting period in winter 2009-2010 were further fitted with CFDD and CTDD, respectively, and the regression equations of sea ice area and sea ice thickness with CFDD and CTDD were obtained, as shown in the following figure: Fig.19 Fig. 20 shown.

[0134] The linear regression equation of sea ice area and CFDD during the freezing period is Y=115.056* X +4634.98, R 2 =0.943, indicating that there is a high correlation between the sea ice area and CFDD during the freezing period of 2009-2010; the linear regression equation between the sea ice area and CTDD during the melting period is: Y =-1091.633* X +55871.077, R 2 =0.933. The negative correlation between sea ice area and CTDD during the melting period is high. The linear regression equation between sea ice thickness and CFDD during the freezing period is: Y =0.00036315* X +0.0838, R 2 =0.535, the linear regression equation between sea ice thickness and CTDD during the melting period is: Y =-0.00453* X +0.26534, R 2 =0.786. It can be seen that the growth and ablation process of sea ice is closely related to the accumulation of temperature, and different accumulated temperature index models are suitable for sea ice in the freezing and melting periods.

[0135] Since the area of ​​a certain sea is large and the development degree of winter sea ice in the three major bays is different, the present invention selects CFDD and CTDD with the highest correlation among meteorological factors to conduct regional research on the certain sea.

[0136] Table 6 Correlation coefficients between freezing and melting periods and temperature factors in Bay 1, Bay 2 and Bay 3

[0137] Note: ** indicates significant correlation at the 0.01 level (two-tailed); * indicates significant correlation at the 0.05 level (two-tailed).

[0138] The sea was divided into four major areas, and the correlation analysis between sea ice and meteorological factors (CFDD, CTDD) was conducted for Bay 1, Bay 2, and Bay 3. As shown in Table 6, during the total ice period, the sea ice area of ​​Bay 1 had the highest correlation with CFDD (r=0.519), and the sea ice thickness of Bay 1 had the highest correlation with CFDD (r=0.251); during the freezing period, the sea ice area of ​​Bay 3 had the highest correlation with CFDD (r=0.967), and the sea ice thickness of Bay 1 had the highest correlation with CFDD (r=0.537); during the melting period, the sea ice area of ​​Bay 3 had the highest correlation with CTDD (r=-0.781), and the sea ice thickness of Bay 3 had the highest correlation with CTDD (r=-0.267).

[0139] 3.2 The impact of wind field on the sea ice conditions in a certain sea. A certain sea is affected by the edge effect of the Siberian-Mongolian high pressure in winter. The present invention uses CFSR wind field data to calculate the wind field distribution in a certain sea in the winter of 2003~2023. It can be seen that from December of each year to January and February of the following year, the wind speed in a certain sea tends to weaken from strong. In December and January, the wind direction is mainly northwest, and it turns to the north wind direction after entering February. In general, the wind speed distribution in different months in winter generally has a distribution trend of low nearshore and high farshore. At the same time, the change of wind speed in the longitudinal direction is more obvious than that in the latitudinal direction. The wind speed in a certain bay one is significantly higher than that in a certain bay two and a certain bay three. In December, the wind speed in the coastal waters of a certain sea is about 5 m / s, and the closer to the center, the stronger the wind speed. The maximum wind speed area is in the middle of a certain bay, with a wind speed of about 8 m / s; the overall distribution of wind speed in January of the following year is similar to that in December, with the characteristics of lower wind speed in coastal waters than in central waters and wind speed contours roughly parallel to the coastline. The wind speed is lower than that in December, with wind speeds of 3~4m / s in the coastal waters of a certain sea and 6~8m / s in the central waters of a certain sea; the wind speed continues to decrease in February, and the wind speed decreases in areas above 6 m / s; the average wind direction in the wind field throughout the winter is northwest, and the wind field distribution is characterized by low wind speed near the coast and high wind speed far from the coast, and the wind speed is roughly parallel to the coastline.

[0140] Taking the Bohai wind field in the winter of 2009-2010 when the ice situation was severe as an example, compared with the wind field in the winter of a certain sea from 2003 to 2022, the overall wind speed in the winter of a certain sea in 2009 was larger. The wind direction in January 2010 was basically the same as that in December 2009, with prevailing northwesterly winds. Compared with the average wind speed distribution in January 2003-2022, the area with wind speeds above 6 m / s in a certain sea in January 2010 was larger, and the wind speeds in the central sea area of ​​a certain bay and the central sea area of ​​a certain sea were above 8 m / s. Compared with December 2009 and January 2010, the wind direction in February 2010 changed significantly, with the wind direction being northward and the wind speed decreasing. The wind speed in a certain bay was 7-8 m / s, and the wind speed in the central sea area of ​​a certain sea was above 8 m / s. The correlation analysis between the daily average wind speed in winter from 2003 to 2022 and the sea ice area in a certain sea was carried out, and the correlation coefficient between the sea ice area and the average wind speed in a certain sea was -0.190, indicating that the two are negatively correlated but not strongly related.

[0141] 3.3 The impact of salinity on the sea ice conditions in a certain sea. The salinity of seawater is an important factor affecting the characteristics of sea ice. The salinity data of the present invention comes from the three-dimensional temperature-salinity inversion field data provided by the National Marine Science Data Center of China. The data uses the Northwest Pacific Ocean Real-time Analysis System to convert the ocean surface data into sub-surface temperature-salinity data, and uses the real-time / quasi-real-time data of international public and central observation and monitoring to assimilate and correct the reconstructed field to provide accurate three-dimensional temperature-salinity field data, and realize the real-time analysis of ocean surface information to underwater three-dimensional temperature-salinity field. After comparing with the measured salinity data of a certain bay in December 2006, it is consistent with the actual temporal and spatial distribution. The salinity data used in the present invention is the salinity data 5 m below sea level, which is the salinity of seawater under the ice layer.

[0142] The salinity data of a certain sea area from December 1 to March 10 every year from 2016 to 2021 were selected for processing to obtain the average salinity of a certain sea, as shown in the figure. Fig.21 The daily average salinity of a certain sea area from 2016 to 2021 shows that the salinity of a certain sea drops to the lowest point at the end of December and the beginning of January each year. After the sea ice area increases, the salinity of seawater also rises rapidly. As the sea ice area increases, the salinity of seawater increases. As the thickness of the ice layer increases, the growth of sea ice slows down, and the ice needles are regularly arranged vertically and directional. Salt is easier to discharge, which increases the salinity of seawater. Therefore, the end of December and the beginning of January in winter is the early stage of sea ice development. The development speed is fast and the thickness of sea ice is small. Salinity is easier to discharge, and the salinity of seawater increases faster.

[0143] 3.4 The impact of SSC on the sea ice conditions in a certain sea. The MODIS L1B data was processed into L2-level remote sensing reflectance (Rrs) data, and then the data was preprocessed by geometric correction, atmospheric correction and other preprocessing. The inversion model of suspended sediment in the surface layer of a certain sea used in the present invention was established by Zhou Zhou, and the SSC inversion formula is: ; In the formula, is the concentration of suspended matter in seawater, is the natural index, is the reflectivity of the Rrs555 band; the correlation between the 555nm band and the surface suspended matter concentration of a certain sea is 0.63 using this formula, and the SSC concentration obtained by inversion based on the remote sensing reflectivity of the optical signal is not accurate, but the present invention mainly studies the changing trend of the surface suspended matter concentration of a certain sea in winter, which does not affect the research conclusion of the present invention.

[0144] The reflectivity of the Rrs555 band cannot reflect the concentration characteristics of suspended solids on the surface of the sea area covered by sea ice, so the remote sensing reflectivity is de-iced. Shi et al. pointed out that the short-wave infrared band reflectivity of the ocean surface covered by sea ice will decrease significantly, and based on this, a sea ice discrimination model was established. When the data reflectivity meets the following four conditions, it is judged as sea ice: ; ; ; ; In the formula, is the normalized off-water reflectance, is the normalized off-water reflectance at 859 nm. is the normalized off-water reflectance at 555 nm. is the normalized off-water reflectance in the 412nm band, is the remote sensing reflectivity.

[0145] The average SSC distribution in a certain sea area in winter from 2003 to 2020, unit: mg / L, the high-value area of ​​suspended matter concentration is roughly in the modern Yellow River Delta and the northern part of a certain bay. The high-value range of SSC gradually expands from December to February. In December, the high-value area of ​​SSC in a certain sea is mainly on the south coast of a certain bay two and the west coast of a certain bay three; in January of the following year, strong northerly winds disturbed the sea surface water, causing more intense sediment resuspension, which increased the SSC in the sea area near the Yellow River estuary. At the same time, the suspended matter diffused significantly outward from the Yellow River estuary. The high-value area of ​​SSC in February did not change much from January, but there was a trend of decreasing in the east-west direction, and the SSC in the coastal waters decreased significantly. Overall, the high-value area of ​​SSC in a certain sea in winter is mainly in the sea area near the Yellow River estuary. The reason for the high SSC in a certain bay two and a certain bay three is mainly due to the large amount of sediment resuspension at the mouth of the Yellow River. The SSC of a certain bay one is significantly lower than that of a certain bay three and a certain bay two.

[0146] The correlation coefficients between the sea ice level and the monthly average SSC and the winter average SSC are 0.50, 0.68, 0.40 and 0.56, respectively, indicating that the ice level in a certain sea area has a significant positive correlation with SSC. However, it is worth noting that the correlation coefficients between the monthly average SSC and the winter average SSC and the wind speed in the same period are 0.20, 0.65, 0.38 and 0.40, respectively, indicating that the influence of wind speed on SSC is also positively correlated. Years with severe ice conditions are often accompanied by a sharp drop in temperature and strong cold air, resulting in large winds and waves during the sea ice development period, which strongly disturbs the sea water and leads to a high SSC. From Jiang Man's in-situ observation results, it can be seen that the presence of sea ice will lead to a decrease in wave-induced bed shear stress but an increase in flow-induced bed shear stress. Since the flow-induced bed shear stress has little effect on the sediment resuspension process, the coverage of sea ice weakens the sediment resuspension, thereby reducing SSC to a certain extent.

[0147] Taking the winter of 2009-2010 as an example, the influence of sea ice and SSC in a certain sea is analyzed. The average distribution of suspended sediment concentration in each month of the winter of 2009-2010 is obtained after de-icing. The distribution law of SSC in the three major bays in the winter of 2009-2010 is consistent with the average SSC distribution in many years. Compared with the suspended sediment concentration in a certain sea in the winter of 2003-2022, the SSC in the winter of 2009-2010 is higher, and the SSC distribution range of the high-value area is larger, which is particularly significant in January. The reason for this phenomenon may be related to the stronger wind speed in the winter of 2009-2010, which caused more intense sediment resuspension. The daily average SSC distribution of the MODIS observation image date close to the sea ice period in the winter of 2009-2010 after de-icing. The SSC value is larger in the area near the ice edge. The reason for this phenomenon may be that the water flow is hindered by sea ice when passing through sea ice-covered waters and open waters, and eddies and other phenomena are formed, stirring suspended matter into nearby waters and increasing SSC.

[0148] 4. Return period of sea ice thickness in a certain sea.

[0149] 4.1 Processing of inversion results: Due to the large sea area and the large number of pixels in the MODIS observation images, which will affect the image processing speed, the inversion results are pre-processed before calculating the sea ice thickness reappearance value.

[0150] (1) Eliminate outliers. Due to the presence of thin clouds over a certain sea area, snow covering the sea ice surface, and other obstructions, the previous preprocessing work cannot completely remove the obstructions. Therefore, there are a small number of pixels in the inversion data results where the sea ice thickness inversion results are too large. Therefore, it is necessary to eliminate or smooth the outliers. The present invention eliminates outliers in the inversion results based on the sea ice observation data and other measured data published in the "China Marine Disaster Bulletin". First, the pixel value is compared with the difference with other adjacent pixels. If it exceeds the actual observed thickness of sea ice in previous years, it is marked as an outlier. Extreme values ​​and outliers are eliminated by smoothing the image using a spatial filter; (2) Resampling. The spatial resolution of MODIS L1B data is 1 km. Since the area of ​​a certain sea is large, the amount of pixel data is too large to affect the processing speed. Therefore, the present invention chooses to divide the sea area into 5×5 km grids, uses the grid center coordinates to represent the pixel position, and selects the maximum value of sea ice thickness within the set grid as the grid pixel value;

[0151] (3) Convert the inverted and resampled data from 2009 to 2022 into ASCII format data corresponding to the coordinates and sea ice thickness values.

[0152] 4.2 Probability distribution theory and parameter estimation. The time scale of the sea ice data of the present invention is the winter of 2003~2022, which is a long time scale. The sea ice return period of the present invention is the estimation of the annual maximum value of sea ice thickness. The Pearson-Ⅲ model is more suitable for longer time series, and the use of Pearson-Ⅲ can avoid the trouble of processing extremely large values ​​and selecting fitting parameters. Based on the research of previous scholars on the calculation of the return period model, the present invention selects the Pearson-Ⅲ type to estimate the return period of sea ice thickness in a certain sea. The Pearson distribution family contains a total of 13 curves, among which the third curve Pearson-Ⅲ is more effective in hydrological analysis. The Pearson-Ⅲ distribution is also called the Γ distribution, and its probability density function and distribution function are as follows: ; ; In the formula, is the frequency density function, for of Distribution function, is a random variable, is a positional parameter, is the shape parameter, is the scale parameter, is the probability distribution function, It is a multi-year maximum value. is the natural index; Assume that the annual maximum value of the sea ice thickness index is The probability density function of , its distribution function for: ; In the formula, is the probability distribution function; ; Annual maximum value The probability of occurrence is the guaranteed rate , and its calculation formula is: In the formula, Probability guarantee rate, .

[0153] 4.3 Distribution of sea ice in a certain sea with a multi-year return period, such as Fig. 22As shown. Based on the maximum value of sea ice thickness in the long-term series from 2003 to 2022, the spatial distribution of ice thickness under extreme conditions of sea ice in a certain sea with different return periods is calculated. The overall sea ice thickness of a certain sea with a return period of one year is less than 0.3m. There is a small area of ​​sea ice with a thickness of 0.25~0.3m along the coast of a certain bay, the southern coast of a certain bay, and the coast of a certain bay. Sea ice with a thickness of 0.1~0.25m is mainly distributed in the northern 1 / 2 of a certain bay. The thickness of sea ice in the eastern part of a certain bay is higher than that in the western part. In a certain bay, a certain bay, and a certain bay, sea ice with a thickness of 0.1~0.25m is mainly developed along the coastline, and the thickness of sea ice in the central sea area is below 0.05m. The maximum sea ice thickness of a 5-year return period is about 0.6m, and the area where it exists is similar to the distribution of sea ice thickness of a 1-year return period. Thicker sea ice appears in the coastal areas of the three major bays and the northern coast of a certain sea area. The thickness of sea ice in a certain bay is mainly between 0.2 and 0.5 m. The closer to the center of a certain sea, the smaller the thickness is, and the thickness on the west coast is smaller than that on the east coast. The 0.25 m sea ice thickness contour line is almost parallel to a certain coastline. From the spatial distribution of sea ice thickness in a 10-year return period, it can be seen that the sea ice thickness continues to increase, and the 0.25 m sea ice thickness contour line is significantly closer to the center of a certain sea area. Even in a dense sea ice area in a certain bay, there is a very small area of ​​sea ice with a thickness of about 0.75 m. In the 10-year return period medium ice thickness distribution, the closer the sea ice is to the shore, the thicker the sea ice is. The sea ice thickness in a certain bay is mainly concentrated between 0.3 and 0.5 m, the sea ice thickness in a certain bay two and a certain bay three is concentrated between 0.15 and 0.3 m, and the sea ice thickness in the coastal area is about 0.4 to 0.5 m. The thickness of the 25-year sea ice in Bay 1 is mainly 0.2-0.6 m, and the 0.25 m sea ice thickness contour line further develops towards the central area of ​​Bay 1. The range of sea ice with a thickness of more than 0.75 m increases, mainly concentrated in the northern coast and eastern coast of Bay 1 and the dense sea ice area of ​​Bay 1. The distribution range of thicker sea ice in Bay 2 and Bay 3 did not increase significantly, but the sea ice thickness range in this area increased to 0.25-0.35 m, but the maximum sea ice thickness distribution range increased, and sea ice with a thickness of more than 0.75 m appeared in the northern and southern coasts of Bay 3. From the 50-year sea ice thickness, we can see that the 0.1 m sea ice thickness contour and the 0.25 m sea ice thickness contour are not much different from the 25-year sea ice thickness, but the overall sea ice thickness has further increased, the sea ice area with a thickness of more than 0.75 m has increased, and the area has increased. Sea ice with a thickness of more than 0.75 m mainly appears in the northern coast of Bay 1 and the dense sea ice area of ​​Bay 1. In addition, sea ice with a thickness of more than 0.75 m also develops in the southern coast of Bay 3. In the 50-year sea ice thickness distribution, a small area of ​​sea ice with a thickness of more than 1 m appeared in the dense sea ice area of ​​Bay 1 and the southern coast of Bay 3.

[0154] Overall, the sea ice area and thickness of Bay 1 in different return periods are higher than those of Bay 2 and Laizhou. The ice situation in the eastern part of Bay 1 is significantly heavier than that in the western part of Bay 1, and the ice thickness values ​​are larger in the northern and southern waters of Bay 3. Due to the combined influence of right-hand tidal currents, coastal currents and northerly winds, when floating ice or broken ice appears on the sea surface of Bay 1, it will accumulate on the east and southeast coasts of Bay 1, forming most of the drift ice concentrated in the east and southeast. At the same time, after the remnant of the Yellow Sea warm current enters the sea area of ​​Bay 1, it first flows to the western coast of Bay 1, causing the sea water temperature in the west of Bay 1 to be higher than that in the east. Therefore, the ice situation in the eastern part of Bay 1 is significantly greater than that in the western part of Bay 1. Since the winter monsoon in Bay 1 is dominated by northerly winds, under the influence of the rightward Coriolis force in the northern hemisphere, sea ice is more likely to accumulate on the southern coasts of Bay 3 and Bay 2. According to the measured data, the thickness of sea ice accumulation in the Zhuangxi Sea Area reached more than 0.8 m in 2009, and the thickness of sea ice accumulation in the sea area near the Qingdong Artificial Island reached 2 m in 2013. In addition, by using the Landsat 8 satellite with higher spatial resolution to observe the sea area along the three coasts of a certain bay, it was found that there are a large number of aquaculture ponds along the three coasts of a certain bay with shallow waters and prone to thick ice.

[0155] It can be seen from the above embodiments that the present invention inverts the sea ice thickness in winter from 2003 to 2022 based on MODIS L1B optical remote sensing data, and extracts the sea ice area of ​​a certain sea in previous years in combination with Sentinel-1 A / B data in winter from 2015 to 2022, and analyzes the generation and disappearance laws and spatial distribution of the sea ice thickness and sea ice area of ​​a certain sea; combines ERA5 reanalysis data, CSFR wind field data, and salinity data to conduct a correlation analysis on the factors affecting the growth and development of sea ice (air temperature, sea surface temperature, cumulative freezing degree days CFDD, cumulative melting degree days CTDD, wind field, and salinity). In addition, MODIS L1B data is used to invert the SSC of a certain sea in winter, and the mutual influence between the SSC of a certain sea in winter and sea ice is analyzed. Finally, the spatial distribution of sea ice thickness under different return periods is obtained based on the Pearson-Ⅲ probability distribution model. The following results are obtained: (1) The spatial distribution of sea ice in a certain sea is mainly affected by the depth of sea water. The main ice-forming areas of Bay 1 are concentrated in the sea area with a water depth of less than 20m, and the ice-forming areas of Bays 2 and 3 are mainly in the sea area with a water depth of less than 10m along the southeast coast; the spatial distribution of sea ice thickness in a certain sea follows the rule that the thickness is larger near the coast and smaller offshore, and there is no ice formation in the central sea area of ​​a certain sea; the overall sea ice thickness on the east coast of Bay 1 is greater than that on the west coast, and the sea ice thickness of Bays 2 and 3 is relatively large, mostly concentrated in the southern and southeastern coastal waters; (2) During the freezing period, the meteorological factor with the highest correlation with sea ice area and sea ice thickness is CFDD (r=0.844, r=0.599); on the melting day, the meteorological factor with the highest correlation with sea ice area and sea ice thickness is CTDD (r=-0.524, r=-0.248); among the three bays of a certain sea, during the total ice period, the sea ice area of ​​a certain bay has the highest correlation with CFDD (r=0.519), and the sea ice thickness of a certain bay has the highest correlation with CFDD (r=0.251); during the freezing period, the sea ice area of ​​a certain bay has the highest correlation with CFDD (r=0.967), and the sea ice thickness of a certain bay has the highest correlation with CFDD (r=0.537); during the melting period, the sea ice area of ​​a certain bay has the highest correlation with CTDD (r=-0.781), and the sea ice thickness of a certain bay has the highest correlation with CTDD (r=-0.267); (3) The thickness of sea ice in a certain sea was inverted by combining the MODIS optical image data of the past 20 years with the measured sea ice thickness data, and the maximum value of sea ice thickness in winter from 2003 to 2022 was obtained. The annual extreme value data of ice thickness were statistically analyzed using a 5×5km grid to form a data set of maximum ice thickness of grid points over the years. The spatial distribution of ice thickness with a return period of 5 years, 10 years, 25 years and 50 years for each grid point was calculated using the Pearson-Ⅲ probability extreme value distribution function. The results show that the sea ice in the coastal waters of a certain sea is serious, the northern part of a certain bay is more serious than the southern part, the eastern part is more serious than the western part, and the ice situation in the southeast of a certain bay and a certain bay is serious, which is consistent with the actual sea ice distribution characteristics.

[0156] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0157] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A remote sensing-based method for assessing factors affecting sea ice conditions. It is characterized in that The method includes: S1, select multi-year winter sea MODIS L1B data, extract multi-year winter sea ice area time series, use multi-year winter Sentinel-1A / B data based on gray level co-occurrence matrix calculation to extract sea ice texture features and identify sea ice types, and then supplement optical observation sea ice images; S2, using MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, the sea ice thickness is inverted to obtain the multi-year winter sea ice thickness and spatial distribution map, and the spatial distribution law of sea ice is analyzed; S3, using CSFR reanalysis wind field data and ERA5 reanalysis sea surface temperature data, and the three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center, obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice condition level; S4, based on the sea ice thickness inversion data, the Pearson-Ⅲ distribution model is used to obtain the spatial distribution of extreme sea ice thickness under different return periods.

2. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S1, multi-year winter sea MODIS L1B data are selected and multi-year winter sea ice area time series are extracted, including: In optical imaging sea ice monitoring, the surface directional reflectivity is converted into spectral albedo, and the narrow-band reflectivity is converted into wide-band reflectivity; Narrow band reflectivity Defined as the upward radiation flux from a surface and downward radiated energy Ratio of: ; ; In the formula, is the multi-band reflectivity, is the multi-band upward radiation flux, is the downward radiation flux of the band, is the solar zenith angle, Band range , They are the minimum and maximum wavelengths of a certain band, respectively, the visible light band The range is 0.3μm-0.7μm; use The three reflectivity values ​​are estimated and the obtained parameters are used to calculate the irradiance. Under the condition that the surface reflection conforms to Lambert's reflection law, the upward irradiance of the earth's surface and the downward irradiance of the earth's surface are: ; ; In the formula, is the irradiance without surface reflection, is the surface reflectivity, is the atmospheric hemispherical albedo, is the solar zenith angle, is the apparent downward irradiance, To observe the zenith angle, is the downlink irradiance, The path of the sun, is the atmospheric transmittance of the observation path; According to different reflectivity values, we can get Input the variable value, get the surface downlink irradiance and surface reflected irradiance, get the MODIS data narrow band albedo information, and use the MODIS data narrow band albedo and wide band conversion formula: ; In the formula, is the wide-band albedo of MODIS data, They are the narrow-band albedo corresponding to MODIS bands 1-5 and 7; For the inversion of sea ice thickness, Grenfell and Allison obtained a conversion model between wide-band albedo and sea ice thickness based on optical remote sensing theory. The formula is: ; In the formula, is the reflectivity of the spectrum of solar shortwave radiation, is the sea ice reflectivity corresponding to infinite ice thickness, is the correlation coefficient, , is the seawater albedo, is the thickness of sea ice, is the correlation coefficient of reflectivity, is the natural exponential function; Take 1.209, Taking 0.7, the root mean square error between the inverted sea ice thickness and the measured thickness is 0.

02.

3. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S1, multi-year winter Sentinel-1A / B data are used to extract sea ice texture features and identify sea ice types based on gray-level co-occurrence matrix operations, including: S101, using GLCM to obtain eight texture feature parameters of Sentinel-1 A / B images, including mean grayscale, variance, contrast, uniformity, dissimilarity, entropy, angular second moment, and correlation; the selected GLCM parameters include window size, direction angle, displacement, and grayscale quantization level; S102, sea ice sample collection, using the spectral characteristics of sea ice to identify ice skin, gray ice and white ice in winter sea areas, extracting sea ice types as samples from the ice skin, gray ice, white ice and water reflectance in the VV polarization mode of the ocean in Sentinel-1 A / B images; S103, classify sea ice based on a combination of multiple texture features. Based on the gray-level co-occurrence matrix (GLCM) method suitable for extracting SAR image texture parameters, select the optimal sea ice feature texture combination and the optimal classification method. Use four easily identifiable samples to classify and identify sea ice. Count the areas of sea water, ice skin, gray ice and white ice in the area. Classification and identification methods include support vector machine classification (SVM), neural network classification method, and maximum likelihood supervised classification (MLC).

4. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S2, the sea ice thickness is inverted using MODIS L1B data based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model to obtain the multi-year winter sea ice thickness and spatial distribution map, and analyze the sea ice spatial distribution law, including: The winter sea ice conditions are divided into: light ice year, normal ice year obviously lighter, normal ice year slightly lighter, normal ice year slightly lighter, normal ice year, normal ice year slightly heavier, normal ice year heavier, normal ice year obviously heavier and heavy ice year; We analyzed the ice conditions in the years with heavy ice in the sea area, the ice conditions in different winter years, and the temporal and spatial distribution characteristics of sea ice. We used MODIS L1B optical remote sensing satellite data and Sentinel-1A / B SAR image data as data sources to classify and identify sea ice in the sea area in multi-year winters, analyze the development and change trends of multi-year sea ice, and obtain the maximum thickness of sea ice along the coast through the spatial distribution of the maximum value of multi-year sea ice thickness. Analyze the trend of multi-year sea ice development and change, including selecting statistics on the frequency and area of ​​sea ice in multi-year winters. The definition of the frequency of sea ice is: ; In the formula, is the probability of a pixel being frozen in the winter of a certain year, is the number of days in the remote sensing image where the pixel has ice formation in this winter. is the total number of this pixel in the remote sensing image in this winter.

5. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S3, obtaining the relationship between air temperature and sea ice and ice condition level includes: Two parameters are introduced to measure the contribution of temperature to the development and melting of sea ice and its duration: cumulative frozen degree days CFDD and cumulative melt degree days CTDD, which represent the cumulative value below or above a critical temperature value over a period of time. The calculation formula is as follows: ; ; In the formula, For time, For the end time, is the start time, is the critical temperature for sea ice development and melting. ; It is the average daily temperature of the sea area on that day.

6. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S3, obtaining the relationship between the atmospheric wind field and the sea ice and ice condition level includes: The CFSR wind field data is used to calculate the wind field distribution in the sea area in winter for many years, including the calculation of the total average wind field distribution in December, January, February and winter for many years.

7. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S3, obtaining the relationship between salinity and sea ice and ice condition level includes: The salinity data in the sea area are selected for processing to obtain the average salinity of the sea area, including the calculation of the daily average salinity in December, January and February for many years.

8. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S3, it also includes obtaining the relationship between the sea surface suspended matter concentration SSC and the sea ice and ice condition level, processing the MODIS L1B data into L2 level remote sensing reflection Rrs data, and performing geometric correction and atmospheric correction preprocessing on the data; the SSC inversion formula is: ; In the formula, is the concentration of suspended matter in seawater, is the natural index, is the reflectivity of the Rrs555 band; When the reflectivity meets the following four conditions, it is judged as sea ice: ; ; ; ; In the formula, is the normalized off-water reflectance, is the normalized off-water reflectance at 859 nm. is the normalized off-water reflectance at 555 nm. is the normalized off-water reflectance in the 412nm band, is the remote sensing reflectivity.

9. The method for evaluating sea ice conditions influencing factors based on remote sensing according to claim 1, It is characterized in that In step S4, based on the inversion data of sea ice thickness, the Pearson-III distribution model is used to obtain the spatial distribution of extreme sea ice thickness under different return periods, including: S401, inversion result processing, including removing outliers, resampling, and converting the inversion and resampling data of each year for many years into ASCII format data corresponding to coordinates and sea ice thickness values; S402, probability distribution theory and parameter estimation, choose Pearson-Ⅲ type to estimate the sea ice thickness return period; Pearson-Ⅲ type distribution is also called Γ type distribution, and the probability density function and distribution function are as follows: ; ; In the formula, is the frequency density function, for of Distribution function, is a random variable, is a positional parameter, is the shape parameter, is the scale parameter, is the probability distribution function, It is a multi-year maximum value. is the natural index; Assume that the annual maximum value of the sea ice thickness index is The probability density function of , its distribution function for: ; In the formula, is the probability distribution function, It is a maximum value that occurs once in many years; ; Annual maximum value The probability of occurrence is the guaranteed rate The calculation formula is: In the formula, is the probability guarantee rate, .

10. A remote sensing-based assessment system for sea ice conditions influencing factors. It is characterized in that The system is used to evaluate the remote sensing-based sea ice condition impact factor evaluation method according to any one of claims 1 to 9, and the system comprises: The sea ice type recognition module is used to select multi-year winter sea MODIS L1B data and extract the multi-year winter sea ice area time series. It uses multi-year winter Sentinel-1A / B data to extract sea ice texture features and identify sea ice types based on gray-level co-occurrence matrix operations, thereby supplementing optical observation sea ice images. The module for obtaining the sea ice thickness and spatial distribution map of a certain sea in multi-year winter is used to use MODIS L1B data, based on the narrow-band and wide-band mutual conversion formula and the wide-band albedo and sea ice thickness relationship model, to invert the sea ice thickness, obtain the multi-year winter sea ice thickness and spatial distribution map, and analyze the spatial distribution law of sea ice; The module for obtaining the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level is used to obtain the relationship between air temperature, sea surface temperature, wind field, sea surface suspended matter concentration and salinity and sea ice and ice level by using CSFR reanalysis wind field data, ERA5 reanalysis sea surface temperature data and three-dimensional temperature-salinity field inversion data provided by the Ocean Science Data Center. The module for obtaining the spatial distribution of extreme sea ice thickness in a certain sea under different return periods is used to obtain the spatial distribution of extreme sea ice thickness under different return periods based on the sea ice thickness inversion data and using the Pearson-Ⅲ distribution model.

Citation Information

Patent Citations

  • Remote sensing estimation method for sea ice surface accumulated snow thickness

    CN112504144A

  • High-precision sea ice concentration inversion method

    CN116030362A

  • Sea ice daily data comprehensive processing method and device and computing equipment

    CN116597319A

  • Sea ice condition data reconstruction method and device based on multi-source remote sensing image

    CN118037818A

  • Ice thickness estimation method by SAR

    JP2005291782A

Cited By

  • Sea ice information processing method, device and equipment based on remote sensing image

    CN121392631A

  • Sea ice thickness inversion method based on multi-source heterogeneous fusion and physical constraint

    CN122174202A