Meteorological drought remote sensing monitoring method and system based on small water body transfer matrix

Abstract

The invention discloses a meteorological drought remote sensing monitoring method and system based on a small water body transfer matrix, and belongs to the field of remote sensing drought monitoring. Adopting a random forest algorithm and a space-time neighborhood similarity algorithm to supplement the extracted half-moon-scale surface water to obtain a half-moon-scale water body product; dividing the half-month-scale water body product into three states based on the area size, and performing pairwise space matching to obtain six water body state conversion types; selecting two types of converting a large water body into a small water body and converting a non-water body into the small water body, and calculating NSWSTI by combining the total transfer times of the two types; and carrying out meteorological drought remote sensing monitoring in combination with the traditional drought index and the NSWSTI. According to the method, the problem of water body data missing caused by cloud layer coverage is effectively solved, the dynamic change of the small water body in the drought process is more sensitively captured, and the precision and reliability of remote sensing drought monitoring are improved.

Description

Technical Field

[0001] The invention belongs to the field of remote sensing drought monitoring, and in particular relates to a meteorological drought remote sensing monitoring method and system based on a small water body transfer matrix. Background Art

[0002] Traditional drought monitoring mainly relies on data obtained from traditional ground monitoring stations, which are very limited in terms of spatial continuity. In addition, traditional methods have problems such as low monitoring accuracy, long monitoring cycle, and limited spatial coverage, and are often unable to fully and accurately grasp the occurrence and evolution of drought events. Therefore, it is of great significance to explore new drought monitoring methods that can be promoted in similar basins in China and even the world by utilizing the characteristics of remote sensing technology such as continuous monitoring time and convenient acquisition of monitoring data, combined with the transformation of surface water states that is most directly related to drought.

[0003] Drought is generally divided into four types: meteorological drought refers to a period when rainfall is lower than normal and temperature is higher than normal; agricultural drought refers to a period when soil moisture is lower than average rainfall or higher than evaporation rate; hydrological drought refers to a period when reservoir water level is low and river flow is lower than average. The interaction between these three types of drought can lead to ecological and socioeconomic drought. Scientists have proposed different drought monitoring models and methods, including monitoring soil moisture using thermal inertia models and retrieving soil moisture using active / passive microwave remote sensing methods. However, in order to further quantify the severity of drought, determine the evolution process of drought, and study the distribution characteristics of drought, many scholars at home and abroad have gradually developed many drought indices. The most widely used drought indices include the standardized precipitation index (SPI), the standardized runoff index (SRI), the standardized precipitation evapotranspiration index (SPEI), the standardized soil moisture index (SSMI), and the Palmer Drought Severity Index (PDSI). These indices mainly assess the severity of drought based on the hydrological balance between precipitation and evaporation, taking into account the relative relationship between precipitation and evaporation to reveal the changing trend of soil moisture. In addition, these indices are usually standardized to eliminate regional and temporal differences and summarize them as deviations from the long-term average.

[0004] Although the above-mentioned traditional drought indices integrate multiple factors such as precipitation, evaporation, and soil moisture, they still mainly rely on ground observation data. Specifically, drought indices are usually calculated using data from meteorological stations, but the spatial distribution of meteorological stations is limited and cannot fully cover the monitoring area. Due to the uneven distribution of meteorological stations, especially when monitoring large areas or areas with complex terrain, monitoring based solely on meteorological station data is prone to limitations and cannot accurately reflect the spatial variability of drought in the entire region. Accurate results are often not obtained in terms of regional homogeneity. Although the use of remote sensing technology to incorporate surface water dynamics into drought monitoring can achieve spatial continuity, it still faces many challenges. First, the water body is quantified only in terms of area, and there is a lack of mechanistic explanation for the dynamics of surface water during the evolution of drought; second, the existing products are incomplete due to cloud cover, and it is difficult to accurately capture small water bodies.

[0005] Therefore, it is necessary to design and develop a meteorological drought remote sensing monitoring method and system based on small water body transfer matrix to address the above problems. Summary of the invention

[0006] The purpose of the present invention is to address the problems existing in the prior art and to provide a meteorological drought remote sensing monitoring method and system based on a small water body transfer matrix, which not only effectively solves the problem of water body data missing due to cloud cover, improves the accuracy and reliability of remote sensing drought monitoring, but also more accurately reflects the occurrence, development and spatial expansion characteristics of drought, providing important technical support for regional water resources management and ecological environmental protection.

[0007] According to one aspect of the present invention, a method for remote sensing monitoring of meteorological drought based on a small water body transfer matrix is ​​provided, comprising:

[0008] Based on the acquired DW water body data, semi-lunar scale surface water is extracted. Combined with the collected auxiliary water body data, the random forest algorithm and spatiotemporal neighborhood similarity algorithm are used to supplement the extracted semi-lunar scale surface water to obtain a semi-lunar scale water body product.

[0009] The half-moon scale water body products are divided into three states based on the area size, and six water body state conversion types are obtained by pairwise spatial matching based on the three states. The total number of transfers of each of the six water body state conversion types is calculated to form a state transfer matrix;

[0010] Select two types: large water body turning into small water body and non-water body turning into small water body, and calculate the normalized small water body state transition index NSWSTI by combining the total number of transitions from large water body to small water body and from non-water body to small water body;

[0011] A correlation analysis is performed by combining the traditional drought index and the normalized small water body state transition index NSWSTI, and meteorological drought remote sensing monitoring is performed based on the results obtained from the correlation analysis.

[0012] Furthermore, combined with the collected auxiliary water body data, the random forest algorithm and spatiotemporal neighborhood similarity algorithm are used to supplement the extracted semi-lunar scale surface water, including:

[0013] The auxiliary water body data collected include Sentinel-1 SAR images, Sentinel-2 optical remote sensing images, digital elevation model data, and maximum surface water range data;

[0014] The water body data of the Sentinel-1 SAR image is extracted by using a random forest algorithm, and the missing area of ​​the semi-lunar scale surface water is supplemented to obtain an initial semi-lunar scale water body product;

[0015] For Sentinel-2 optical remote sensing images, the index threshold method is used to mask the desert and ice and snow areas in the initial half-moon scale water products;

[0016] The initial half-moon scale water product is further supplemented with water using the spatiotemporal neighborhood similarity algorithm;

[0017] Using the mountain shadow layer data obtained from the digital elevation model data, the mountain shadow caused by the high terrain in the study area is masked to obtain the final half-moon scale water product.

[0018] Furthermore, the index threshold method also includes The calculation formula is as follows:

[0019]

[0020] in, is the normalized water index, is the reflectance value of the green band of the image, is the reflectance value in the near-infrared band; 7.

[0021] Furthermore, the half-moon scale water products are divided into three states based on the size of the area, including:

[0022] Water area ≥ 10 5 square meters, recorded as large water bodies;

[0023] Water area <10 5 Square meters and not connected to large rivers are recorded as small water bodies;

[0024] Other land use types except water bodies are recorded as non-water bodies.

[0025] Furthermore, the half-moon scale water products are divided into three states based on the area size, including:

[0026] Small water bodies were masked using maximum surface water extent data.

[0027] Furthermore, the formula for calculating the normalized small water body state transition index NSWSTI is:

[0028]

[0029] in, is the total number of transfers from large water bodies to small water bodies, It is the total number of transfers from non-water bodies to small water bodies.

[0030] According to one aspect of the present invention, a remote sensing drought monitoring system based on a small water body transfer matrix is ​​provided, comprising:

[0031] The module for obtaining water products is used to extract semilunar-scale surface water based on the acquired DW water data, and to obtain semilunar-scale water products by combining the collected auxiliary water data, the collected Sentinel-1 Synthetic Aperture Radar (Sentinel-1 SAR) images, Sentinel-2 optical remote sensing images and Digital Elevation Model (DEM) data, and using the random forest algorithm and the spatiotemporal neighborhood similarity algorithm to supplement the extracted semilunar-scale surface water.

[0032] A state transition module is used to divide the semi-moon scale water body product into three states based on the area size, perform pairwise spatial matching based on the three states to obtain six water body state transition types, and calculate the total number of transfers of each of the six water body state transition types to form a state transition matrix;

[0033] The NSWSTI calculation module is used to select two types of large water bodies turning into small water bodies and non-water bodies turning into small water bodies, and calculate the normalized small water body state transition index NSWSTI by combining the total number of transfers from the large water body to the small water body and from the non-water body to the small water body;

[0034] The drought monitoring module is used to combine the traditional drought index and the normalized small water body state transition index NSWSTI to perform correlation analysis, and based on the results obtained from the correlation analysis, perform meteorological drought remote sensing monitoring.

[0035] According to one aspect of the present invention, there is provided an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the meteorological drought remote sensing monitoring method based on a small water body transfer matrix when executing the computer program.

[0036] According to one aspect of the present invention, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the meteorological drought remote sensing monitoring method based on small water body transfer matrix are implemented.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention proposes a meteorological drought remote sensing monitoring method and system based on small water body transfer matrix, which effectively solves the problem of missing water body data due to cloud cover through Sentinel-1 SAR image and spatiotemporal neighborhood similarity algorithm, and improves the accuracy and reliability of remote sensing drought monitoring.

[0039] 2. The present invention proposes a meteorological drought remote sensing monitoring method and system based on a small water body transfer matrix. Through desert and ice and snow masks and mountain shadow masks, the final half-moon scale water body product is obtained, and further through the division of water body states, a water body state transfer matrix is ​​formed. This can identify water body change signals in the early stage of drought earlier, greatly improving the accuracy of remote sensing drought monitoring.

[0040] 3. The present invention proposes a meteorological drought remote sensing monitoring method and system based on the small water body transfer matrix, which more accurately reflects the occurrence, development and spatial expansion characteristics of drought by proposing the normalized small water body state transition index NSWSTI, and provides important technical support for regional water resources management and ecological environment protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 A schematic flow chart of a meteorological drought remote sensing monitoring method based on a small water body transfer matrix of the present invention;

[0043] Figure 2 It is a flow chart of a water body extraction method in a meteorological drought remote sensing monitoring method based on a small water body transfer matrix of the present invention;

[0044] Figure 3 A schematic diagram of small water body transfer in a meteorological drought remote sensing monitoring method based on a small water body transfer matrix of the present invention;

[0045] Figure 4 The present invention is a schematic diagram of the module structure of a remote sensing drought monitoring system based on a small water body transfer matrix. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, the present invention provides a meteorological drought remote sensing monitoring method based on a small water body transfer matrix, comprising: collecting data, wherein the data includes DW water body data, Sentinel-1 Synthetic Aperture Radar (Sentinel-1 SAR) images, Sentinel-2 optical remote sensing images, Digital Elevation Model (DEM) data, maximum surface water range (JRC Global Surface Water Mapping Layers, JRC) data, and meteorological data; extract semi-lunar scale surface water based on the water category with a label index of 0 in the DW water body data, and use the random forest algorithm and the spatiotemporal neighborhood similarity algorithm to supplement the extracted semi-lunar scale surface water to obtain a semi-lunar scale water body product; divide the semi-lunar scale water body product into three states based on the area size, perform pairwise spatial matching based on the three states to obtain six water body state conversion types, calculate the total number of transfers of the six water body state conversion types, and form a water body state transfer matrix; select the two types of large water bodies turning into small water bodies L_S and non-water bodies turning into small water bodies N_S, which have opposite change trends in theory, and construct a normalized small water body state conversion index NSWSTI based on the total number of transfers from the large water body to the small water body L_S and the non-water body to the small water body N_S; conduct correlation analysis based on the traditional drought index and the normalized small water body state conversion index NSWSTI, and conduct meteorological drought remote sensing monitoring based on the results obtained by the correlation analysis.

[0048] Specifically, meteorological data include the daily precipitation dataset CHM_PRE for China obtained by the National Tibetan Plateau Data Center, other meteorological statistical data such as the Bulletin on Flood and Drought Prevention and the Yearbook of Meteorological Disasters, both from the website of the Ministry of Water Resources of China, and the global high-resolution atmospheric reanalysis dataset ERA5 (the fifth generation of ECMWF Global Climate Atmospheric Reanalysis) developed and maintained by the European Centre for Medium-Range Weather Forecasts ECMWF.

[0049] Specifically, when there is only a single data source of optical images, there is still a problem of incomplete coverage due to cloud cover. The random forest algorithm is used to extract water data from Sentinel-1 SAR images as a supplement to the DW semi-lunar scale surface water, and a preliminary semi-lunar surface water product is obtained.

[0050] Specifically, due to the differences in satellite imaging in different regions and the limitations of the extraction algorithm, there are still many small water bodies that cannot be captured. The spatiotemporal neighborhood similarity algorithm is used to further supplement and finally obtain a complete half-moon surface water product. For the desert ice and snow areas and mountain shadow areas that exist in the initial half-moon scale water product, mask processing is performed.

[0051] Specifically, the half-moon scale water products are divided into three states according to the size of the water area. 5 Square meters is a small water body (SmallWater, S), larger than 10 5 Square meters are large water bodies (LargeWater, L), and other land covers are non-water bodies (NoWater, N). In terms of space, the data of two consecutive time periods are spatially matched, and the water products in the three states are converted into each other to form a state transfer matrix, and six types of water state conversion are obtained.

[0052] Specifically, by analyzing the changing trends of various transfer types in drought, two types of change, large water body to small water body L_S and non-water body to small water body N_S, are selected from the six types of water body state transition. The total number of transitions from large water body to small water body L_S and non-water body to small water body N_S is calculated, and the normalized small water body state transition index NSWSTI is constructed. The calculated total number of transitions is substituted to obtain the NSWSTI result. The NSWSTI calculation formula is as follows:

[0053]

[0054] in, is the total number of transfers from large water bodies to small water bodies, It is the total number of transfers from non-water bodies to small water bodies.

[0055] The embodiment of the present invention selects the Poyang Lake Basin and the Dongting Lake Basin as the data area, wherein the experimental configuration is: the online programming adopts the GEE platform, the language is Javascript; the local programming adopts PyCharm 2024.1.1 with the language as Python, and the configuration environment is Python 2.7, ArcGISx6410.8 and Python 3 anaconda3. The specific implementation steps are as follows:

[0056] (1) When conducting long-term drought monitoring, there may be cloudy, overcast and rainy weather in some areas, which will lead to poor integrity of optical image interpretation of DW water products and a large number of data missing areas. The random forest algorithm is used to extract water data from Sentinel-1 SAR images, and the missing areas of surface water at the semi-lunar scale are supplemented to obtain the initial semi-lunar scale water product. The spatiotemporal neighborhood similarity algorithm is used to further supplement the initial semi-lunar scale water product to obtain the semi-lunar scale water product.

[0057] Specifically, for areas with missing data, the morphological filtering method was used to eliminate the scattering noise of Sentinel-1 SAR images, and five features, namely VV, VH, VV+VH, VV-VH, and VV / VH, were selected as sample features. 2000 water bodies and non-water bodies were manually selected as sample points, and the random forest algorithm was used for classification to obtain the supplementary water bodies of the Sentinel-1 SAR images, which were then merged with the water body data of DW to obtain the initial half-moon scale water body product.

[0058] Specifically, an algorithm based on the similarity of spatiotemporal neighborhoods was used to supplement water bodies in areas where there were still vacancies. According to the results of multiple experiments, gap pixels with a radius of 70 m were selected as the spatial neighborhood, and the similarity threshold was set to 30. The DW surface water data, the supplementary water bodies of Sentinel-1 SAR images, and the supplementary water bodies based on spatiotemporal similarity were merged to obtain the final half-moon scale water product.

[0059] Specifically, for the presence of desert and ice and snow, Perform desert snow masking as follows:

[0060]

[0061] 7

[0062]

[0063] in, is the normalized water index; is the reflectance value of the green band of the image; is the reflectance value in the near-infrared band.

[0064] For some areas with higher terrain, in order to eliminate the influence of mountain shadow, the mountain shadow layer obtained from DEM data is used for masking.

[0065] (2) In the early stages of a drought, precipitation decreases, and shallow, small, and self-regulating water bodies gradually disappear. As the drought progresses, precipitation continues to decrease, and larger water bodies may break up into multiple small water bodies. If the drought continues to worsen, these small water bodies will continue to disappear. When the drought eases and precipitation gradually increases, shallow water will form again, and adjacent small water bodies may merge into a complete large water body.

[0066] Based on the dynamic changes in water bodies caused by the above drought, it is determined that 10 5 Square meters is the critical value for water body status classification. The area of ​​a single water body is greater than 10 5 Square meters is a large water body, less than 10 5 Square meters are small water bodies, and other terrestrial vegetation are non-water bodies; these three states are converted into each other to form a water state transfer matrix. The definition is as follows:

[0067]

[0068]

[0069] in, is the current transfer times matrix in grid i, is the total number of transfers in the basin during the current period, L_S is the conversion of large water bodies to small water bodies, and N_S is the conversion of non-water bodies to small water bodies. Taking into account the area of ​​the example area and in order to be consistent with the scale of precipitation data, a 10km×10km grid is used.

[0070] Specifically, the extracted water products were converted into vector data, and each feature was divided into a grid of 10 km and a grid of more than 10 5 The water category attribute of square meters is assigned the value of "LargeWater"; after masking it with the JRC data of the maximum surface water extent, the extracted area is less than 10 5 The water category attribute of each square meter is assigned the value "SmallWater". Special processing is performed on non-water features, that is, the water bodies in 24 time periods of the year are merged to generate the maximum range of the year (MaxWater), and then the surface water and MaxWater of the year are erased, and the category attribute is added and assigned the value "NoWater". It is ensured that the data of each time period contains three water body categories, namely LargeWater, SmallWater and NoWater.

[0071] Specifically, for each grid cell, the closest features of the water body elements in two adjacent time periods are searched to ensure that the water body elements in the two time periods are accurately matched, and then the transfer type attribute is added to the matching features. For example, if the current state is a small water body and the previous state is a large water body, it is represented as L_S.

[0072] Specifically, the six types of transfer within each grid are accumulated and summed one by one, and finally the water state transfer data at the 10km spatial grid scale are obtained. Since the transfer is calculated based on the water products of the previous period and the next period, after the above calculation, a total of 23 transfer matrices can be generated for each year.

[0073] During the drought, large water bodies will gradually split into small water bodies, while during the drought relief period, a large number of small water bodies will be formed and eventually converge into part of the large water body. Therefore, in the entire drought evolution process, the state transition related to small water bodies is particularly critical, such as Figure 3 Through the permutation and combination experiments of six different transfer types, two state transition types with opposite change trends in theory, namely L_S and N_S, were selected according to the above mechanism to construct a new drought index.

[0074] Specifically, in order to highlight the differences between drought development and alleviation and narrow the order of magnitude differences between different transfer types, L_S and N_S were normalized by calculating the ratio of the difference between L_S and N_S and their sum; in addition, the smaller the drought index value, generally represents a higher degree of drought, and the normalized value is subtracted from 1 to finally obtain the normalized small water body state transition index NSWSTI.

[0075] Specifically, at the grid scale, the index is calculated only when the number of L_S and N_S in the grid is not zero, otherwise, the index is considered empty; at the watershed scale, the sum of the number of L_S and N_S of each grid cell is first accumulated, and then the index value of the entire study area is calculated according to the following formula:

[0076]

[0077]

[0078]

[0079]

[0080] in, and They represent the number of transitions of L_S and N_S in grid cell i respectively; and They represent the total number of conversions of L_S and N_S in the basin, respectively; represents the index value of a particular grid cell, and Represents the index value for the entire watershed.

[0081] (4) Calculate the drought index at different time scales (n months), usually 1, 3, 6, 9 and 12 months. In order to match the NSWSTI results of the semi-monthly scale water body product, the SPI, SRI and SSMI at five scales are calculated, which are recorded as 01, 03, 06, 09 and 12 respectively, where 01 represents the drought index at the semi-monthly scale and 12 represents the drought index at the semi-annual scale.

[0082] Specifically, the Pearson correlation coefficient is a statistical indicator used to measure the degree of linear correlation between two variables X and Y. Its value is between -1 and 1, and is represented by R Value. The closer R is to 1 or -1, the higher the correlation between the two variables. The closer R is to 0, the lower the correlation, or even no correlation. The p value indicates whether the relationship between the two variables is significant, with a certain error probability. If the p value is less than 0.05, the linear relationship between the two variables is considered significant; if the p value is greater than 0.05, the linear relationship between the two variables is considered insignificant.

[0083] Specifically, after comparing the two basins, it was found that the correlation coefficient between the NSWSTI index and various drought indices in the Poyang Lake Basin was higher than that in the Dongting Lake Basin. The drought indices such as SPI, SRI, and SSMI were calculated in combination with the collected meteorological data. As shown in Table 1, the NSWSTI index has the highest correlation with SPI 01, which is 0.52 in the Dongting Lake Basin and 0.54 in the Poyang Lake Basin; the correlation between the NSWSTI index and SRI 01 is 0.35 in the Dongting Lake Basin and 0.46 in the Poyang Lake Basin; the correlation between the NSWSTI index and SSMI is relatively low overall.

[0084] Specifically, the correlation coefficient between the NSWSTI index and the shallow soil moisture index (SSMI1 01) was high, at 0.46 in the Dongting Lake Basin, but gradually decreased with the deepening of the soil layer, reaching only 0.24 in SSMI4 01. These results indicate that small shallow water bodies are highly sensitive to drought changes and are directly related to precipitation. Due to their limited ability to penetrate deeper soil layers, they can effectively reflect precipitation patterns and describe drought conditions.

[0085] Specifically, from the perspective of drought indices at different time scales, the NSWSTI index has the highest correlation with the half-month scale (SPI 01, SRI01 and SSMI1 01). As the scale increases, the correlation decreases and even fails to pass the significance test. In general, drought indices at smaller time scales are more sensitive to meteorological droughts, which fully verifies the important role of small water bodies in short-term droughts. In addition, the NSWSTI index is more sensitive to drought indices at short time scales and is suitable for monitoring meteorological droughts.

[0086] Table 1 Correlation coefficients between NSWSTI and traditional drought indices

[0087] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides a meteorological drought remote sensing monitoring system based on a small water body transfer matrix, which is used to execute a meteorological drought remote sensing monitoring method based on a small water body transfer matrix in the above method embodiment.

[0088] The embodiment of the present invention provides a meteorological drought remote sensing monitoring system based on a small water body transfer matrix, such as Figure 4 As shown in FIG, the specific implementation process includes: extracting semi-lunar scale surface water based on the acquired DW water body data, combining the collected Sentinel-1 SAR images, Sentinel-2 optical remote sensing images and digital elevation model data, using random forest algorithm and spatiotemporal neighborhood similarity algorithm to supplement the extracted semi-lunar scale surface water, and obtaining a semi-lunar scale water body product; then, the semi-lunar scale water body product is divided into three states based on the area size through the state conversion module, and six water body state conversion types are obtained by pairwise spatial matching based on the three states, and the total number of transfers of each of the six water body state conversion types is calculated respectively; in the NSWSTI calculation module, the two types of theoretically opposite change trends, namely, large water body conversion to small water body and non-water body conversion to small water body, are selected, and the result is shown in FIG. The total number of transfers from the large water body to the small water body and from the non-water body to the small water body is combined to construct the normalized small water body state transition index NSWSTI; finally, the drought monitoring module is used to conduct a correlation analysis by combining the traditional drought index and the normalized small water body state transition index NSWSTI. The results show that the drought index on a smaller time scale is more sensitive to meteorological drought, and the NSWSTI index is more sensitive to the drought index on a short time scale. It is suitable for remote sensing monitoring of meteorological drought, and more accurately reflects the occurrence, development and spatial expansion characteristics of drought, providing important technical support for regional water resources management and ecological environmental protection.

[0089] Based on the same inventive concept as the aforementioned embodiment, an embodiment of the present invention also provides an electronic device, including a memory and a processor, the memory being used to store computer-executable instructions, and the processor being used to execute computer-executable instructions, to implement the remote sensing drought monitoring method based on the small water body transfer matrix proposed in the aforementioned embodiment.

[0090] The embodiment of the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by the processor, it is used to identify water body change signals in the early stage of drought earlier, accurately reflect the occurrence, development and spatial expansion characteristics of drought, and improve the accuracy of remote sensing drought monitoring. The storage medium can be any non-volatile storage device such as a hard disk, a solid-state hard disk, a flash drive, an optical disk, etc., for storing computer program codes and necessary data files. The stored computer program includes: a water body product acquisition module, a state conversion module, a NSWSTI calculation module, and a drought monitoring module.

[0091] Finally, it should be pointed out that the above specific embodiments are only representative examples of the present invention. Obviously, the present invention is not limited to the above specific embodiments, and there are many variations. Any simple modification, equivalent changes and modifications made to the above specific embodiments based on the technical essence of the present invention should be considered to belong to the protection scope of the present invention.

Claims

1. A meteorological drought remote sensing monitoring method based on small water body transfer matrix, characterized in that: include: Based on the acquired DW water body data, semi-lunar scale surface water is extracted. Combined with the collected auxiliary water body data, the random forest algorithm and spatiotemporal neighborhood similarity algorithm are used to supplement the extracted semi-lunar scale surface water to obtain a semi-lunar scale water body product. The half-moon scale water body products are divided into three states based on the area size, and six water body state conversion types are obtained by pairwise spatial matching based on the three states. The total number of transfers of each of the six water body state conversion types is calculated to form a state transfer matrix; Select two types: large water body turning into small water body and non-water body turning into small water body, and calculate the normalized small water body state transition index NSWSTI by combining the total number of transitions from large water body to small water body and from non-water body to small water body; A correlation analysis is performed by combining the traditional drought index and the normalized small water body state transition index NSWSTI, and meteorological drought remote sensing monitoring is performed based on the results obtained from the correlation analysis.

2. The meteorological drought remote sensing monitoring method based on small water body transfer matrix according to claim 1 is characterized in that: Combined with the collected auxiliary water body data, the random forest algorithm and spatiotemporal neighborhood similarity algorithm are used to supplement the extracted semi-lunar scale surface water, including: The collected auxiliary water data include Sentinel-1 SAR images, Sentinel-2 optical remote sensing images, digital elevation model data and maximum surface water range data; The water body data of the Sentinel-1 SAR image is extracted by using a random forest algorithm, and the missing area of ​​the semi-lunar scale surface water is supplemented to obtain an initial semi-lunar scale water body product; For Sentinel-2 optical remote sensing images, the index threshold method is used to mask the desert and ice and snow areas in the initial half-moon scale water products; The initial half-moon scale water product is further supplemented with water using the spatiotemporal neighborhood similarity algorithm; Using the mountain shadow layer data obtained from the digital elevation model data, the mountain shadow caused by the high terrain in the study area is masked to obtain the final half-moon scale water product.

3. The meteorological drought remote sensing monitoring method based on small water body transfer matrix according to claim 2 is characterized in that: The index threshold method also includes The calculation formula is as follows: ， in, is the normalized water index, is the reflectance value of the green band of the image, is the reflectance value in the near-infrared band.

4. The meteorological drought remote sensing monitoring method based on small water body transfer matrix according to claim 1 is characterized in that: The half-moon scale water products are divided into three states based on the area size, including: Water area ≥ 10 5 square meters, recorded as large water bodies; Water area <10 5 Square meters and not connected to large rivers are recorded as small water bodies; Other land use types except water bodies are recorded as non-water bodies.

5. The meteorological drought remote sensing monitoring method based on small water body transfer matrix according to claim 4 is characterized in that: The half-moon scale water products are divided into three states based on the area size, including: The small water bodies are masked using the maximum surface water extent data.

6. The meteorological drought remote sensing monitoring method based on small water body transfer matrix according to claim 1 is characterized in that: The formula for calculating the normalized small water body state transition index NSWSTI is: ， in, is the total number of transfers from large water bodies to small water bodies, It is the total number of transfers from non-water bodies to small water bodies.

7. A remote sensing drought monitoring system based on small water body transfer matrix, characterized in that: include: The module for obtaining water products is used to extract semi-lunar scale surface water based on the acquired DW water data. Combined with the collected auxiliary water data, the random forest algorithm and spatiotemporal neighborhood similarity algorithm are used to supplement the extracted semi-lunar scale surface water to obtain semi-lunar scale water products. A state transition module is used to divide the semi-moon scale water body product into three states based on the area size, perform pairwise spatial matching based on the three states to obtain six water body state transition types, and calculate the total number of transfers of each of the six water body state transition types to form a state transition matrix; The NSWSTI calculation module is used to select two types of large water bodies turning into small water bodies and non-water bodies turning into small water bodies, and calculate the normalized small water body state transition index NSWSTI by combining the total number of transfers from the large water body to the small water body and from the non-water body to the small water body; The drought monitoring module is used to combine the traditional drought index and the normalized small water body state transition index NSWSTI to perform correlation analysis, and based on the results obtained from the correlation analysis, perform meteorological drought remote sensing monitoring.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the meteorological drought remote sensing monitoring method based on small water body transfer matrix described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the meteorological drought remote sensing monitoring method based on small water body transfer matrix described in any one of claims 1 to 6 are implemented.

Images