Ice surface snow melting detection method based on intensity ratio

Abstract

The application provides an ice surface snow melting detection method based on intensity ratio, and belongs to the technical field of image processing, and comprises the following steps: (1) calculating the sea ice density of each pixel point; (2) calculating the sea ice contribution brightness temperature of each pixel point at 36.5 GHz and 89.0 GHz respectively; (3) calculating the sea ice contribution polarization ratio of each pixel point at 36.5 GHz and 89.0 GHz; (4) calculating the polarization ratio threshold at 36.5 GHz and 89.0 GHz according to the intensity ratio principle; (5) marking the snow melting points to obtain first snow melting data and second snow melting data; and (6) taking the union of the snow melting points in the first snow melting data and the second snow melting data to obtain final ice surface snow melting detection data. The ice surface snow melting detection method can calculate the sea ice density, extract the sea ice contribution brightness temperature in the radiometer observation mixed pixel, and determine the polarization ratio threshold according to the sea ice contribution polarization ratio, so that the interference of seawater on the brightness temperature in the pixel can be removed, the snow melting point detection precision is improved, and high-precision and all-around monitoring of the snow melting process can be realized.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and more specifically, relates to a method for detecting snow melt on Arctic ice surfaces based on intensity ratio. Background Technology

[0002] In recent years, Arctic sea ice extent has continued to decrease, and snow cover has also shown a trend of decreasing and earlier melting. In spring, snow cover protects the ice surface from solar radiation, delaying the onset of sea ice melting. During the snowmelt season, melting snow forms melt pools, reducing surface albedo and promoting rapid sea ice melting. Therefore, snowmelt directly affects the timing, amount, and speed of sea ice melt, playing a significant role in sea ice reduction. When temperatures rise in late spring, the surface emissivity of meltwater forming in lower latitude Arctic regions approaches 1. After the initial melting, as liquid water content further increases, the snow either continues to increase in humidity or refreezes (mostly at night), strongly affecting microwave signals. If snow humidity continues to increase, its emissivity will approach that of open water, a process that typically occurs before melt pools form. As humidity increases in snow, the emissivity at the snow-air interface increases, leading to a corresponding increase in brightness temperature. If refreezing occurs, the accumulation of snow grains under isothermal deformation conditions will result in very large snow grains, increasing the scattering of microwaves by the snow cover and causing a decrease in brightness temperature. Therefore, the instantaneous snow cover state on the sea ice surface has a strong impact on the observed brightness temperature. Accurate snow and ice information is a crucial factor in determining the water and heat exchange between the polar ocean and the atmosphere, further influencing research on climate change in polar regions and even globally. Furthermore, in the process of retrieving Arctic snow depth using microwave radiometers, the snow depth at melting points cannot be determined with the start of the late spring melting season, leading to poor accuracy in melting point detection. Therefore, accurately and reliably identifying the snow melting state on the sea ice surface is essential.

[0003] Currently, there are three main methods for detecting snow melt using microwave radiometers:

[0004] The first method utilizes the temporal variation threshold of brightness temperature, which determines the melting state by detecting abrupt changes in brightness temperature at a certain point over time. For example, the 30K threshold difference method using the brightness temperature of the 19GHz horizontal polarization channel of the SSM / I is used to detect ice cap melting. This method is only sensitive to the instant at which melting begins. During the continuous melting of snow, the brightness temperature change tends to level off, making this method ineffective in determining the melting state.

[0005] The second method is the diurnal brightness temperature variation method. Ramage et al. utilized the characteristic of microwave radiometers revisiting the same location twice daily, determining snow melting based on the difference in brightness temperature between the two visits. Specifically, when the difference exceeds a certain threshold, it is considered snow melting. Furthermore, several scholars both domestically and internationally have used the diurnal brightness temperature variations of the 36.5 GHz channel of AMSR-E and the 37 GHz channel of SSM / I to determine snow melting in the Arctic and Antarctic. However, in certain periods of summer, snow that melts during the day may not freeze at night, and the diurnal brightness temperature variation method may miss such cases.

[0006] The third method uses multi-band brightness temperature difference or cross-polarization gradient ratio to determine the melting time of snow on sea ice or ice sheets. This method uses brightness temperature differences at different frequencies (e.g., 18 GHz and 36 GHz) or with different polarization patterns (horizontal and vertical) to distinguish between dry and wet snow. However, this method relies heavily on experience or statistics and lacks a physical basis. Summary of the Invention

[0007] In order to solve the technical problem of poor detection accuracy caused by unreasonable judgment threshold settings in existing ice surface snow melting detection methods, this invention proposes an ice surface snow melting detection method based on intensity ratio, which can solve the above-mentioned problems.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0009] A method for detecting snow melt on ice surfaces based on intensity ratio, comprising:

[0010] (1) Calculate the sea ice concentration in the Arctic region based on satellite observation brightness temperature and sea ice concentration algorithm, which represents the percentage of sea ice per unit area;

[0011] (2) Based on satellite observations of brightness temperature and sea ice concentration data, calculate the sea ice contribution brightness temperature at 36.5 GHz and 89.0 GHz for each pixel, including the horizontally polarized sea ice contribution brightness temperature at 36.5 GHz. Vertically polarized sea ice at 36.5 GHz contributes to brightness temperature. Horizontally polarized sea ice contributes brightness temperature at 89.0 GHz And the brightness temperature contributed by vertically polarized sea ice at 89.0 GHz. ;

[0012] (3) According to and The sea ice contribution polarization ratio of each pixel at 36.5 GHz was calculated to obtain the first polarization ratio data. ,according to and The sea ice contribution polarization ratio of each pixel at 89.0 GHz was calculated to obtain the second polarization ratio data. ;

[0013] (4) Calculate the polarization threshold at 36.5 GHz based on the first polarization ratio data, including:

[0014] (41) Calculate the intensity ratio of the first polarization ratio data, including dividing the numerical range of the first polarization ratio data into n sub-intervals according to the data distribution characteristics of the first polarization ratio, and counting the corresponding data volume in each sub-interval to obtain F. i (R36.5), where i takes values ​​from 1 to n, calculate the difference between each element in the sub-interval and its adjacent elements, and count the number of differences greater than the critical value, denoted as N. i (R36.5), where n is a positive integer;

[0015] (42) Calculate the intensity ratio of each sub-interval. ;

[0016] ;

[0017] (43) Compare the intensity ratios corresponding to each sub-interval, and determine the first polarization ratio with the largest intensity ratio as the polarization ratio threshold at 36.5 GHz, which is the first polarization ratio threshold;

[0018] Similarly, the polarization ratio threshold at 89.0 GHz is calculated based on the second polarization ratio data, and this threshold is the second polarization ratio threshold.

[0019] (5) Transfer the first polarization ratio data Each element in the data is compared with the first polarization ratio threshold. Elements below the threshold are marked as snow melting points; otherwise, they are marked as non-snow melting points, thus obtaining the first snow melting data. Similarly, the second polarization ratio data is... Each element in the data is compared with the second polarization ratio threshold to obtain the second snow melt data.

[0020] (6) Take the union of the snow melting points in the first snow melting data and the second snow melting data to obtain the final snow melting detection data of the ice surface.

[0021] In some embodiments, the method for calculating sea ice concentration in step (1) is as follows:

[0022] Let the sea ice concentration C be a cubic function of the polarization brightness temperature difference P:

[0023] ;

[0024] in ~ These are the fitting coefficients, ~ We obtain the following system of four linear equations by solving:

[0025] ;

[0026] ;

[0027] Where P0 is the polarization brightness temperature difference of open water when C approaches 0, and P1 is the polarization brightness temperature difference of sea ice when C approaches 1.

[0028] and These are the vertical polarization brightness temperature and the horizontal polarization brightness temperature observed by satellite, respectively.

[0029] In some embodiments, the calculation method for the brightness temperature contributed by sea ice in step (2) is as follows:

[0030] ;

[0031] Where f represents the frequency, which is 36.5 GHz or 89.0 GHz, and p represents the polarization, which is either horizontal polarization (H) or vertical polarization (V). The original brightness temperature at frequency f and polarization mode p. The brightness temperature of open water at frequency f and polarization mode p.

[0032] In some embodiments, the first polarization ratio data in step (3) The calculation method is as follows:

[0033] ;

[0034] Second polarization ratio data The calculation method is as follows:

[0035] .

[0036] In some embodiments, step (43) further includes removing sub-intervals with less than a set threshold. That is, if the amount of data contained in a sub-interval is less than 0.5% of the total amount of data, all first polarization ratios in the sub-interval are filtered out, the maximum value of the intensity ratios corresponding to the remaining sub-intervals is compared, and one of the first polarization ratios in the sub-interval with the maximum intensity ratio is selected as the polarization ratio threshold at 36.5 GHz.

[0037] In some embodiments, the method for determining the polarization ratio corresponding to the maximum value in step (43) includes:

[0038] Based on the sub-interval corresponding to the determined maximum intensity ratio, the first polarization ratio in the sub-interval is selected as the first polarization ratio threshold.

[0039] In some embodiments, the ice surface snow melting detection method further includes acquiring the brightness temperature of the study area at frequency f and polarization mode p at multiple different times, calculating the first polarization ratio threshold and the second polarization ratio threshold of each group of data, comparing the largest first polarization ratio threshold as the first polarization ratio fixed threshold, comparing the largest second polarization ratio threshold as the second polarization ratio fixed threshold, and skipping step (4) when detecting ice surface snow melting at other adjacent times, and using the first polarization ratio fixed threshold and the second polarization ratio fixed threshold to perform steps (1)-(3), (5)-(6).

[0040] Compared with existing technologies, the advantages and positive effects of this invention are as follows: The ice surface snowmelt detection method based on intensity ratio of this invention calculates sea ice concentration, extracts the brightness temperature contributed by sea ice in the mixed pixels observed by radiometers, and determines the polarization ratio threshold based on the sea ice contribution polarization ratio calculated from the brightness temperature. Finally, this polarization ratio threshold is used to detect and identify snowmelt. This method can remove the interference of seawater on brightness temperature in the pixels observed by satellite radiometers, thereby improving the phenomenon of gradient value reduction caused by seawater influence and improving the detection accuracy of snowmelt points. It can achieve high-precision, all-round monitoring of the snowmelt process, and is not limited to snowmelt determination in areas with high sea ice concentration. At the same time, by utilizing the calculation principle of intensity ratio, threshold determination can be performed for different spatiotemporal ranges, improving the robustness of snowmelt point detection and the portability of other sensors.

[0041] By calculating the intensity ratio of polarization ratio data, the sea ice contribution polarization ratio corresponding to the maximum intensity ratio is found as the polarization ratio threshold. This method utilizes the radiation difference between snowmelt points and non-snowmelt points, where the difference in sea ice contribution polarization ratio is largest near the boundary between the two. According to the definition of intensity ratio, the difference between intensity ratio and sea ice contribution polarization ratio is positively correlated, representing the change of each pixel corresponding to each sea ice contribution polarization ratio with its surrounding values. Therefore, the sea ice contribution polarization ratio with the largest change can be determined by the maximum intensity ratio, and the threshold for judging snowmelt points can be obtained from this. Compared with the existing technology that gives a fixed threshold based on experience or statistics, this method can improve the detection accuracy of snowmelt points.

[0042] Based on the radiation characteristics of sea ice and water, the 36.5 GHz wavelength significantly increases the dielectric constant of liquid water generated by snowmelt, resulting in a marked increase in the detected brightness temperature, enabling precise capture of the phase transition process from dry to wet snow. The brightness temperature difference between vertical and horizontal polarization at 36.5 GHz can distinguish the influence of the underlying surface, improving the accuracy of freeze-thaw discrimination. The 89.0 GHz wavelength, with its shorter wavelength and higher spatial resolution than 36.5 GHz, can capture local snowmelt details and freeze-thaw changes in thin snow-covered areas. It exhibits a strong response to thin melt crusts enriched with surface liquid water, allowing for the detection of early snowmelt and refreezing processes. This scheme achieves high-precision, comprehensive monitoring of the snowmelt process by taking the union of snowmelt points detected using 36.5 GHz brightness temperature data and 89.0 GHz brightness temperature data as the final snowmelt detection result.

[0043] Other features and advantages of the present invention will become clearer after reading the detailed description of the embodiments of the present invention in conjunction with the accompanying drawings. Attached Figure Description

[0044] Figure 1a This is a comparison of simulated brightness temperatures with different liquid water contents in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0045] Figure 1b This is a simulated brightness temperature map of dry snow in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0046] Figure 2a This is an example of the ice surface snow melting detection method based on intensity ratio proposed in this invention, showing the original brightness temperature polarization ratio distribution at 36.5 GHz;

[0047] Figure 2b This is an example of the ice surface snow melting detection method based on intensity ratio proposed in this invention, showing the original brightness temperature polarization ratio distribution at 89.0 GHz;

[0048] Figure 3a This is a distribution map of sea ice contribution polarization ratio at 36.5 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0049] Figure 3b This is a distribution map of sea ice contribution polarization ratio at 89.0 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0050] Figure 4a This is a numerical range and statistical graph of the sea ice contribution polarization ratio at 36.5 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0051] Figure 4bThis is a numerical range and statistical graph of the sea ice contribution polarization ratio at 89.0 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0052] Figure 5 This is an example of the intensity ratio distribution of sea ice contribution polarization ratio at 36.5 GHz and 89.0 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0053] Figure 6a This is an embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention, using the snow melting point distribution map obtained by detecting the sea ice contribution polarization ratio at 36.5 GHz;

[0054] Figure 6b This is an example of the snow melting point distribution map obtained by detecting the sea ice contribution polarization ratio at 89.0 GHz in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention;

[0055] Figure 7 This is one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention. Figure 6a and Figure 6b The effect of merging the snow spots in the middle layer;

[0056] Figure 8 This is a graph showing the threshold variation of the sea ice contribution polarization ratio at two frequencies in one embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention.

[0057] Figure 9 This is a flowchart of an embodiment of the ice surface snow melting detection method based on intensity ratio proposed in this invention. Detailed Implementation

[0058] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] like Figure 1aAs shown, this embodiment uses a MEMLS (Microwave Emission Model of Layered Snowpacks) snow model to simulate the variation of radiation brightness temperature with snow depth when the liquid water content (wet) is 0.1%, 0.5%, 1%, 5%, and 10%. Taking a 36.5GHz vertical polarization channel as an example, Figure 1a The simulated brightness temperature (Wt) for different liquid water contents is shown; for comparison, Figure 1b It shows the simulated brightness temperature under the condition of dry snow.

[0061] Depend on Figure 1a , Figure 1b It is evident that dry and wet snow exhibit significant differences in their radiation characteristics. When the liquid water content (wet) changes from 0 to 0.1%, the surface radiation brightness temperature of the snow increases significantly, and its relationship with snow depth also changes markedly. When the liquid water content (wet) in the snow increases from 0 to 1%, TB (36.5V) rises from 231.7K to 245.9K, an increase of nearly 15K. As the liquid water content (wet) increases, the brightness temperature gradually increases, but the rate of increase slows down. Furthermore, snow containing liquid water hardly changes with snow depth. This is because the scattering of microwaves by wet snow primarily originates from the surface. Therefore, the effect of wet snow on brightness temperature is only related to the surface snow accumulation environment and not to the underlying snow or sea ice. Thus, this method uses brightness temperature data for snow melting detection, and the results conform to actual physical change patterns.

[0062] Because the melting process at the sea ice edge may result in a higher proportion of seawater, the brightness temperature differences at some frequencies may become insignificant. Alternatively, in unmelted sea ice edges, the large proportion of seawater may cause the radiation characteristics within microwave observation pixels to be dominated by seawater, even though the actual snow on the sea ice has not melted, leading to misjudgments. This solution addresses these issues by using horizontal and vertical polarization brightness temperature data at 36.5 GHz and 89.0 GHz.

[0063] Example 1, see Figure 9 As shown in the figure, this embodiment proposes a method for detecting snow melt on ice surfaces based on intensity ratio, including:

[0064] (1) Calculate the sea ice concentration in the Arctic region based on satellite-observed brightness temperature and sea ice concentration algorithm;

[0065] (2) Based on satellite observations of brightness temperature and sea ice concentration data, calculate the sea ice contribution brightness temperature at 36.5 GHz and 89.0 GHz for each pixel, including the horizontally polarized sea ice contribution brightness temperature at 36.5 GHz. Vertically polarized sea ice at 36.5 GHz contributes to brightness temperature. Horizontally polarized sea ice contributes brightness temperature at 89.0 GHz Vertically polarized sea ice at 89.0 GHz contributes to brightness temperature .

[0066] The microwave radiometer onboard the satellite can observe brightness temperature data at different frequency channels. The sea ice contribution brightness temperature of each pixel at 36.5 GHz and 89.0 GHz can be calculated based on the brightness temperature data.

[0067] By using horizontal and vertical polarization brightness temperature data at 36.5 GHz and 89.0 GHz, and based on the radiation characteristics of sea ice and water, the 36.5 GHz wavelength significantly increases the dielectric constant of liquid water generated by snowmelt, resulting in a marked increase in the detected brightness temperature, thus accurately capturing the phase transition process of snow from dry to wet. The difference between vertical and horizontal polarization brightness temperatures at 36.5 GHz can distinguish the influence of the underlying surface, improving the accuracy of freeze-thaw discrimination. The 89.0 GHz wavelength is shorter and has higher spatial resolution than 36.5 GHz, enabling it to capture local snowmelt details and freeze-thaw changes in thin snow cover areas. It also has a strong response to thin melt crusts enriched with surface liquid water, allowing for the detection of early snowmelt and refreezing processes. This scheme achieves high-precision, comprehensive monitoring of the snowmelt process by taking the union of snowmelt points detected based on 36.5 GHz brightness temperature data and those detected based on 89.0 GHz brightness temperature data as the final snowmelt detection result.

[0068] (3) According to and The sea ice contribution polarization ratio of each pixel at 36.5 GHz was calculated to obtain the first polarization ratio data. ,according to and The sea ice contribution polarization ratio of each pixel at 89.0 GHz was calculated to obtain the second polarization ratio data. .

[0069] This scheme uses only the brightness temperature contributed by sea ice and determines the polarization ratio threshold based on the polarization ratio contributed by sea ice calculated from the brightness temperature. Finally, the polarization ratio threshold is used to detect and identify the threshold of snow melting. This can remove the interference of seawater on the brightness temperature and improve the phenomenon of gradient value reduction caused by the influence of seawater.

[0070] (4) Calculate the polarization threshold at 36.5 GHz based on the first polarization ratio data, including:

[0071] (41) Calculate the intensity ratio of the first polarization ratio data, including dividing the numerical range of the first polarization ratio data into n sub-intervals according to the numerical distribution characteristics of the first polarization ratio, and counting the corresponding data in each sub-interval to obtain F. i(R36.5), where i takes values ​​from 1 to n, calculate the difference between each element in the sub-interval and its adjacent elements, and count the number of differences greater than the critical value, denoted as N. i (R36.5), where n is a positive integer.

[0072] By calculating the intensity ratio of polarization ratio data, the sea ice contribution polarization ratio corresponding to the maximum intensity ratio is found as the polarization ratio threshold. This utilizes the fact that there is a radiation difference between snowmelt points and non-snowmelt points, and the difference in sea ice contribution polarization ratio is the largest near the boundary between the two. According to the definition of intensity ratio, the difference between intensity ratio and sea ice contribution polarization ratio is positively correlated, which characterizes the change of the pixel corresponding to each sea ice contribution polarization ratio with the surrounding values. Therefore, the sea ice contribution polarization ratio with the largest change can be determined by the maximum intensity ratio, and the threshold for judging snowmelt points can be obtained from this, thus improving the detection accuracy of snowmelt points.

[0073] (42) Calculate the intensity ratio of each sub-interval. :

[0074] .

[0075] (43) Compare the intensity ratios corresponding to each polarization ratio. The first polarization ratio with the largest intensity ratio is taken as the polarization ratio threshold at 36.5 GHz.

[0076] Similarly, the polarization ratio threshold at 89.0 GHz is calculated based on the second polarization ratio data, and this threshold is the second polarization ratio threshold.

[0077] Specifically, the polarization ratio threshold at 89.0 GHz is calculated based on the second polarization ratio data, including:

[0078] Calculating the intensity ratio of the second polarization ratio data involves dividing the numerical range of the second polarization ratio data into n sub-intervals with a set step size based on the numerical distribution characteristics of the second polarization ratio, and statistically analyzing the corresponding data volume in each sub-interval to obtain F. i (R89.0), where i takes values ​​from 1 to n, calculate the difference between each element in the sub-interval and its adjacent elements, and count the number of differences greater than the critical value, denoted as N. i (R89.0), where n is a positive integer.

[0079] Calculate the intensity ratio of each sub-interval :

[0080] .

[0081] By comparing the intensity ratios corresponding to each polarization ratio, the second polarization ratio with the largest intensity ratio is taken as the polarization ratio threshold at 89.0 GHz, which is the second polarization ratio threshold.

[0082] (5) Transfer the first polarization ratio data Each element in the data is compared with the first polarization ratio threshold. Elements below the threshold are marked as snow melting points; otherwise, they are marked as non-snow melting points, thus obtaining the first snow melting data. Similarly, the second polarization ratio data is... Each element in the data is compared with the second polarization ratio threshold to obtain the second snow melt data.

[0083] (6) To reduce the number of missed detections, the snow melting points in the first snow melting data and the second snow melting data are combined to obtain the final snow melting detection data of the ice surface.

[0084] This embodiment uses satellite microwave brightness temperature data from May 15, 2011 as an example to compare the distribution of various parameters on that day. The original brightness temperature polarization ratio distribution diagrams for the two frequencies are shown below. Figure 2a , Figure 2b As shown, it can be seen that The changes are more pronounced in the peripheral areas. The values ​​are more concentrated.

[0085] The figure only includes points with sea ice concentration greater than 15%, while points with low seawater and sea ice concentrations are not included as they are not very relevant.

[0086] The method for calculating sea ice concentration in step (1) is as follows:

[0087] Let the sea ice concentration C be a cubic function of the polarization brightness temperature difference P:

[0088] ;

[0089] in ~ These are the fitting coefficients, ~ We obtain the following system of four linear equations by solving:

[0090] ;

[0091] ;

[0092] Where P0 is the polarization brightness temperature difference of open water when C is close to 0, and P1 is the polarization brightness temperature difference of sea ice when C is close to 1.

[0093] and These are the vertical polarization brightness temperature and the horizontal polarization brightness temperature observed by satellite, respectively.

[0094] The polarization brightness temperature difference P represents the total polarization brightness temperature difference across various environments, including sea ice, open water, and mixed ice-water regions, and is calculated from brightness temperatures observed by satellite. The polarization brightness temperature difference P(C) containing only sea ice can be expressed as:

[0095] .

[0096] in, The polarization brightness difference representing sea ice, The polarized brightness temperature difference represents the open water area, and C represents the sea ice concentration.

[0097] When C approaches 0, the polarization brightness temperature difference P0 in open water is expressed as: .

[0098] When C approaches 1, the polarization brightness temperature difference P1 of sea ice is expressed as: .

[0099] in and These are the atmospheric influence parameters when C approaches 0 and C approaches 1, respectively. P0 and P1 are called the system point values ​​of the algorithm, and are set as constants in this embodiment. Approaching means infinitely close in mathematics, that is, the difference between C and the approached value is less than the set value. In extreme cases, it can be set to be equal to the approached value.

[0100] Expanding the above formula using Taylor series at C=0 and C=1 respectively, removing higher-order terms, and assuming that atmospheric influence remains constant in all sea ice-covered areas or open waters, we obtain:

[0101] When C→0.

[0102] When C→1.

[0103] then:

[0104] When C→0.

[0105] When C→1.

[0106] In this embodiment, by setting the sea ice concentration C as a cubic function of the polarization brightness temperature difference P, high-precision, physically consistent, and computationally stable inversion can be achieved across the entire concentration range. The calculation method for the sea ice contribution brightness temperature in step (2) is as follows:

[0107] .

[0108] Where f represents the frequency, which is 36.5 GHz or 89.0 GHz, and p represents the polarization, which is either horizontal polarization (H) or vertical polarization (V). This represents the original brightness temperature at frequency f and polarization mode p. The brightness temperature of an open water body at frequency f and polarization mode p is usually expressed as a constant.

[0109] In some embodiments, the first polarization ratio data in step (3) The calculation method is as follows:

[0110] .

[0111] Second polarization ratio data The calculation method is as follows:

[0112] .

[0113] like Figure 3a , Figure 3b The figure shows the distribution of sea ice contribution polarization ratios for the two frequencies mentioned above. It can be seen from the figure that using only the polarization ratio of sea ice contribution brightness temperature eliminates the interference from seawater, and in low-density areas, the values ​​are relatively... Figure 2a , Figure 2b There have been some changes, and the reduction in gradient values ​​caused by the influence of seawater has been improved. Therefore, this scheme determines whether snow cover has melted based on the distribution of the polarization ratio contributed by sea ice.

[0114] Depend on Figure 3a , Figure 3b It can be seen that, regardless of the judgment parameter, the numerical distribution of snow melting points and non-snow melting points is within a certain range, rather than a specific value, and there is some overlap. Furthermore, the distribution of the overall spectral ratio varies depending on the region. Therefore, this scheme also includes setting a threshold to effectively determine the snow melting point.

[0115] This scheme employs the intensity ratio method, the physical meaning of which lies in the fact that, due to the difference in radiation between the snowmelt point and the non-snowmelt point, its... The values ​​differ, and near the boundary point between the two, The difference in values ​​is greatest. If The points with the greatest difference in values ​​are concentrated around a certain value; that value should be the threshold between the two. According to the definition of intensity ratio, it precisely characterizes each... The value corresponds to a point that changes relative to its surrounding values; therefore, based on this ratio, the point with the largest change can be determined. The value is used to obtain the snow melting point. Determining the threshold. Based on this principle, studies were conducted on the contribution of sea ice polarization ratio. and The strength ratio of the two parameters was calculated.

[0116] Step (43) also includes removing sub-intervals with less than a set threshold amount of data. That is, if the amount of data contained in a sub-interval is less than 0.5% of the total amount of data, it is determined that the original amount of data is too small and does not have the basis for calculating the intensity ratio. The corresponding intensity ratio value does not have physical meaning. Therefore, all first polarization ratios in the sub-interval are removed, and the maximum value of the intensity ratios corresponding to the other sub-intervals is compared. One of the first polarization ratios in the sub-interval with the maximum intensity ratio is used as the polarization ratio threshold at 36.5 GHz.

[0117] In some embodiments, the method for determining the polarization ratio corresponding to the maximum value in step (43) includes:

[0118] Based on the determined maximum intensity ratio, the corresponding polarization ratio sub-interval is determined, and the first polarization ratio data is selected as the first polarization ratio threshold.

[0119] Similarly, the second polarization ratio threshold is obtained by processing in the same way as described above.

[0120] The numerical ranges and statistical data of the sea ice contribution polarization ratios for the two frequencies are as follows: Figure 4a , Figure 4b As shown in the figure, Distribution range is relatively large Wider and more concentrated. Based on the numerical distribution ranges of the two, the numerical distribution ranges are set as follows: : 0.8-1, step size set to 0.005; The step size is 0.7-1, with a step size of 0.005. Based on the definition of intensity ratio, the intensity ratio distribution of the two types of sea ice contributing polarization ratios for that day is calculated as follows: Figure 5 As shown. Combined with Figure 4a , Figure 4b and Figure 5 It can be seen that, Figure 5 Because the number of points corresponding to the polarization ratio on both sides of the curve is very small, the intensity ratio of the molecule appears high, but it has no physical meaning. Therefore, in the calculation formula, when the amount of data contained in a sub-interval is less than 0.5% of the total data, the intensity ratio of that interval is not included in the threshold judgment. Apart from the intensity ratios corresponding to these meaningless high-value points on both sides of the data, as the polarization ratio increases, the intensity ratio gradually increases, reaching a peak at around 0.915. The two polarization ratio cases are basically the same.

[0121] Using sea ice contribution polarization ratio respectively and The distribution map of snow melting points obtained from the detection is as follows Figure 6a , Figure 6bAs shown, red pixels represent snow-melted areas, and white pixels represent unmelted areas. It can be seen that due to the difference in the numerical distribution range of the two polarization ratios, the snow-melted areas calculated with similar thresholds differ significantly. The values ​​are generally above 0.95. Therefore, Figure 6a , Figure 6b This can demonstrate the use of the strength ratio method. Differentiating snow melt is not as effective as While it can only distinguish obvious melting areas, it is helpful for detail depiction. Therefore, to avoid missing melting points, this embodiment uses the union of the melting points from both methods to determine the final melting range. The melting detection map after merging the melting points is shown below. Figure 7 As shown.

[0122] To verify the stability of the ice surface snowmelt detection method in this embodiment, the threshold value of the snowmelt polarization ratio was calculated daily in May 2011, and the change of the threshold value over time was observed. The threshold changes of the sea ice contribution polarization ratio at two frequencies are shown below. Figure 8 As shown, the two polarization ratio thresholds, except for May 2nd (where the FY3B / MWRI brightness temperature data was found to be incorrect), are uniformly distributed between 0.9 and 0.93. The values ​​are mainly concentrated between 0.9 and 0.92; The threshold values ​​are mainly concentrated between 0.91 and 0.93, indicating a relatively uniform distribution. Therefore, this scheme adopts a fixed threshold method. Considering the special characteristics of snow melt markers, and to reduce the risk of missed detections, the threshold is set to its maximum value, i.e. Set it to 0.925, and Set it to 0.93.

[0123] In some embodiments, the ice melt detection method further includes acquiring the brightness temperature of the study area at frequency f and polarization mode p at multiple different times, calculating the first polarization ratio threshold and the second polarization ratio threshold for each group of data, comparing the largest first polarization ratio threshold as the first fixed polarization ratio threshold, comparing the largest second polarization ratio threshold as the second fixed polarization ratio threshold, and skipping step (4) when detecting ice melt at other adjacent times, and using the first fixed polarization ratio threshold and the second fixed polarization ratio threshold to execute steps (1)-(3), (5)-(6). This method ensures detection accuracy while reducing computational load.

[0124] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for detecting snow melt on ice surfaces based on intensity ratio, characterized in that, include: (1) Calculate the sea ice concentration in the Arctic region based on satellite observation brightness temperature and sea ice concentration algorithm, which represents the percentage of sea ice per unit area; (2) Based on satellite observations of brightness temperature and sea ice concentration data, calculate the sea ice contribution brightness temperature at 36.5 GHz and 89.0 GHz for each pixel, including the horizontally polarized sea ice contribution brightness temperature at 36.5 GHz. Vertically polarized sea ice at 36.5 GHz contributes to brightness temperature. Horizontally polarized sea ice contributes brightness temperature at 89.0 GHz And the brightness temperature contributed by vertically polarized sea ice at 89.0 GHz. ; (3) According to and The sea ice contribution polarization ratio of each pixel at 36.5 GHz was calculated to obtain the first polarization ratio data. ,according to and The sea ice contribution polarization ratio of each pixel at 89.0 GHz was calculated to obtain the second polarization ratio data. ; (4) Calculate the polarization threshold at 36.5 GHz based on the first polarization ratio data, including: (41) Calculate the intensity ratio of the first polarization ratio data, including dividing the numerical range of the first polarization ratio data into n sub-intervals according to the data distribution characteristics of the first polarization ratio, and counting the corresponding data volume in each sub-interval to obtain F. i (R36.5), where i takes values ​​from 1 to n, calculate the difference between each element in the sub-interval and its adjacent elements, and count the number of differences greater than the critical value, denoted as N. i (R36.5), where n is a positive integer; (42) Calculate the intensity ratio of each sub-interval. ; ； (43) Compare the intensity ratios corresponding to each sub-interval, and determine the first polarization ratio with the largest intensity ratio as the polarization ratio threshold at 36.5 GHz, which is the first polarization ratio threshold; Similarly, the polarization ratio threshold at 89.0 GHz is calculated based on the second polarization ratio data, and this threshold is the second polarization ratio threshold. (5) Transfer the first polarization ratio data Each element in the data is compared with the first polarization ratio threshold. Elements below the threshold are marked as snow melting points; otherwise, they are marked as non-snow melting points, thus obtaining the first snow melting data. Similarly, the second polarization ratio data is... Each element in the data is compared with the second polarization ratio threshold to obtain the second snow melt data. (6) Take the union of the snow melting points in the first snow melting data and the second snow melting data to obtain the final snow melting detection data of the ice surface.

2. The method for detecting snow melting on ice surface based on intensity ratio according to claim 1, characterized in that, The method for calculating sea ice concentration in step (1) is as follows: Let the sea ice concentration C be a cubic function of the polarization brightness temperature difference P: ； coefficient ~ We obtain the following system of four linear equations by solving: ； ； Where P0 is the polarization brightness temperature difference of open water when C approaches 0, and P1 is the polarization brightness temperature difference of sea ice when C approaches 1. and These are the vertical polarization brightness temperature and the horizontal polarization brightness temperature observed by satellite, respectively.

3. The method for detecting snow melting on ice surface based on intensity ratio according to claim 2, characterized in that, The calculation method for the brightness temperature contribution of sea ice in step (2) is as follows: ； Where f represents the frequency, which is 36.5 GHz or 89.0 GHz, and p represents the polarization, which is either horizontal polarization (H) or vertical polarization (V). The original brightness temperature at frequency f and polarization mode p. The brightness temperature of open water at frequency f and polarization mode p.

4. The method for detecting snow melting on ice surface based on intensity ratio according to claim 3, characterized in that, The first polarization ratio data in step (3) The calculation method is as follows: ； Second polarization ratio data The calculation method is as follows: 。 5. The method for detecting snow melting on ice surface based on intensity ratio according to claim 1, characterized in that, Step (43) also includes removing sub-intervals with less than a set threshold. That is, if the amount of data contained in a sub-interval is less than 0.5% of the total amount of data, all first polarization ratios in the sub-interval are removed. The maximum value of the intensity ratios corresponding to the other sub-intervals is compared, and one of the first polarization ratios in the sub-interval with the maximum intensity ratio is selected as the polarization ratio threshold at 36.5 GHz.

6. The method for detecting snow melting on ice surface based on intensity ratio according to claim 5, characterized in that, The method for determining the polarization ratio corresponding to the maximum value in step (43) includes: Based on the sub-interval corresponding to the determined maximum intensity ratio, the first polarization ratio in the sub-interval is selected as the first polarization ratio threshold.

7. The method for detecting snow melting on ice surface based on intensity ratio according to any one of claims 1-6, characterized in that, The ice surface snow melting detection method further includes acquiring the brightness temperature of the study area at frequency f and polarization mode p at multiple different times, calculating the first polarization ratio threshold and the second polarization ratio threshold of each group of data, comparing the largest first polarization ratio threshold as the first polarization ratio fixed threshold, comparing the largest second polarization ratio threshold as the second polarization ratio fixed threshold, and skipping step (4) when detecting ice surface snow melting at other adjacent times, and using the first polarization ratio fixed threshold and the second polarization ratio fixed threshold to perform steps (1)-(3), (5)-(6).

Images