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

Abstract

The application discloses a kind of meteorological drought remote sensing monitoring method and system based on small water body transfer matrix, belong to remote sensing drought monitoring field, it includes: based on the DW water body data extracted half-month scale ground water obtained, combined with the auxiliary water body data collected, using random forest algorithm and space-time neighborhood similarity algorithm to the extracted half-month scale ground water is supplemented, obtains half-month scale water body product;Half-month scale water body product is based on area size and divided into three states, two two space matching obtains six water body state conversion types;Selecting large water body to small water body and non-water body to small water body these two types, combined with the total number of times of the two types of transfer calculation NSWSTI;Combined with traditional drought index and the NSWSTI, carries out meteorological drought remote sensing monitoring.The application effectively solves the problem of water body data loss caused by cloud cover, more sensitively captures the dynamic change of small water body in the drought process, improves the accuracy and reliability of remote sensing drought monitoring.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of remote sensing drought monitoring, and particularly relates to a meteorological drought remote sensing monitoring method and system based on small water body transfer matrix. BACKGROUND

[0002] Traditional drought monitoring mainly relies on data obtained from traditional ground monitoring stations, which is very limited in spatial continuity. In addition, the traditional method has problems such as low monitoring accuracy, long monitoring period, limited spatial coverage, etc., and often cannot comprehensively and accurately grasp the occurrence and evolution process of drought events. Therefore, it is of great significance to explore a new drought monitoring method that can be popularized in China and even similar basins in the world by using remote sensing technology to monitor time continuity, convenient data acquisition, combining with the most direct surface water state transition related to drought.

[0003] Drought is generally divided into four types: meteorological drought refers to the period when rainfall is lower than normal and temperature is higher than normal; agricultural drought refers to the period when soil moisture is lower than average rainfall or higher than evaporation rate; hydrological drought refers to the period when reservoir water level is low and river flow is lower than average. The interaction between the three types of droughts will lead to ecological and socio-economic droughts. Scientists have proposed different drought monitoring models and methods, including monitoring soil moisture using thermal inertia model and retrieving soil moisture using active / passive microwave remote sensing method. 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 standardized precipitation index (SPI), standardized runoff index (SRI), standardized precipitation evapotranspiration index (SPEI), standardized soil moisture index (SSMI), and Palmer drought severity index (PDSI). These indices mainly evaluate the severity of drought based on the hydrological balance between precipitation and evaporation, considering the relative relationship between precipitation and evaporation to reveal the trend of soil moisture. In addition, these indices are usually standardized to eliminate regional and temporal differences, and are summarized as deviations from long-term average values.

[0004] Although the above-mentioned traditional drought indices integrate precipitation, evaporation, soil moisture and other factors, they still mainly rely on ground observation data. Specifically, the drought indices are usually calculated using data of weather stations, and the spatial distribution of weather stations is limited and cannot completely cover the monitoring area. Due to the uneven distribution of weather stations, especially when monitoring large areas or complex terrain areas, monitoring only by relying on weather station data is easily limited and cannot accurately reflect the spatial variability of drought in the whole region, and often cannot obtain accurate results in terms of regional homogeneity. While using remote sensing technology to include surface water dynamics into drought monitoring can achieve spatial continuity, it still faces many challenges. First, only the area of water bodies is quantified, and there is a lack of mechanistic explanation of the dynamic state of surface water in the evolution process of drought; second, 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 transition matrix to solve the above problems. SUMMARY

[0006] The purpose of the present application is to solve the problems existing in the prior art, provide a meteorological drought remote sensing monitoring method and system based on small water body transition matrix, which not only effectively solves the problem of missing water body data caused by 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 resource management and ecological environment protection.

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

[0008] Based on the obtained DW water body data, semi-monthly scale surface water is extracted, and the extracted semi-monthly scale surface water is supplemented by using a random forest algorithm and a spatio-temporal neighborhood similarity algorithm in combination with collected auxiliary water body data, to obtain semi-monthly scale water body products;

[0009] The semi-monthly scale water body products are divided into three states based on the area size, and six water body state transition types are obtained by spatial matching of the three states in pairs, and the total number of transitions of the six water body state transition types is calculated respectively to form a state transition matrix;

[0010] The two types of large water body to small water body and non-water body to small water body are selected, and the normalized small water body state transition index NSWSTI is calculated in combination with the total number of transitions of the large water body to small water body and the non-water body to small water body;

[0011] The correlation analysis is performed in combination with a traditional drought index and the normalized small water body state transition index NSWSTI.

[0012] Further, the extracted semi-monthly scale surface water is supplemented by combining collected auxiliary water body data, using a random forest algorithm and a spatio-temporal neighborhood similarity algorithm, including:

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

[0014] The random forest algorithm is used to extract water body data of the Sentinel-1 SAR images, and the missing areas of the semi-monthly scale surface water are supplemented to obtain an initial semi-monthly scale water body product.

[0015] The exponential threshold method is used for the Sentinel-2 optical remote sensing images to perform mask processing on the desert and ice and snow regions in the initial semi-monthly scale water body product.

[0016] The spatio-temporal neighborhood similarity algorithm is used to further supplement water bodies in the initial semi-monthly scale water body product.

[0017] The mountain shadow layer data obtained from the digital elevation model data is used to perform mask processing on the mountain shadow generated due to the terrain height in the research region to obtain a final semi-monthly scale water body product.

[0018] Further, the exponential threshold method further includes The calculation formula is as follows:

[0019]

[0020] wherein, NWI is a normalized water index, Rg is a reflection value of a green band of an image, Rnir is a reflection value of a near-infrared band; and 7.

[0021] Further, the semi-monthly scale water body product is divided into three states based on the area size, including:

[0022] Water body area ≥ 10 5 square meters, which is recorded as a large water body.

[0023] Water body area < 10 5 square meters and not connected to large rivers and large rivers, which is recorded as a small water body.

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

[0025] Further, the semi-monthly water body product is divided into three states based on the area size, and further includes:

[0026] The small water body is masked using the maximum surface water range data.

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

[0028]

[0029] wherein, is the total number of transitions from large water bodies to small water bodies, is the total number of transitions from non-water bodies to small water bodies.

[0030] According to an aspect of the present application, a remote sensing drought monitoring system based on a small water body transition matrix is provided, comprising:

[0031] The water body product module is used to extract semi-monthly surface water based on the obtained DW water body data, and to supplement the extracted semi-monthly surface water using a random forest algorithm and a spatiotemporal neighborhood similarity algorithm in combination with collected auxiliary water body data, Sentinel-1 Synthetic Aperture Radar (Sentinel-1 SAR) images, Sentinel-2 optical remote sensing images, and Digital Elevation Model (DEM) data, to obtain a semi-monthly water body product.

[0032] The state transition module is used to divide the semi-monthly water body product into three states based on the area size, to obtain six water body state transition types by spatial matching of the three states, to calculate the total number of transitions of each of the six water body state transition types, and to form a state transition matrix.

[0033] The NSWSTI calculation module is used to select two types of large water bodies transitioning to small water bodies and non-water bodies transitioning to small water bodies, to calculate the normalized small water body state transition index NSWSTI in combination with the total number of transitions of the large water bodies transitioning to small water bodies and the non-water bodies transitioning to small water bodies.

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

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

[0036] According to an aspect of the present application, a computer readable storage medium is provided, storing a computer program, wherein the computer program implements the steps of the method for remote sensing monitoring of meteorological drought based on small water body transition matrix when executed by a processor.

[0037] Compared with the prior art, the present application has the following advantages:

[0038] 1. The method and system for remote sensing monitoring of meteorological drought based on small water body transition matrix proposed by the present application effectively solves the problem of missing water body data caused by cloud cover through Sentinel-1 SAR image and spatio-temporal neighborhood similarity algorithm, and improves the accuracy and reliability of remote sensing drought monitoring.

[0039] 2. The method and system for remote sensing monitoring of meteorological drought based on small water body transition matrix proposed by the present application further improves the accuracy of remote sensing drought monitoring by obtaining the final half-month scale water body product through desert and snow cover mask and mountain shadow mask, and forming a water body state transition matrix by dividing the water body state, which can identify the water body change signal in the early stage of drought more quickly.

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

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0042] Figure 1 A flowchart of a method for remote sensing monitoring of meteorological drought based on small water body transition matrix according to the present application;

[0043] Figure 2 A flowchart of a water body extraction method in a method for remote sensing monitoring of meteorological drought based on small water body transition matrix according to the present application;

[0044] Figure 3 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 diagram is a module structure diagram of a remote sensing drought monitoring system based on a small water body transfer matrix of the present invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts 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 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 two types of large water bodies converted to small water bodies L_S and non-water bodies converted to small water bodies N_S with 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 of the correlation analysis.

[0048] Specifically, the meteorological data includes the daily precipitation dataset CHM_PRE of China obtained by the National Tibetan Plateau Scientific Data Center, other meteorological statistical data such as the Flood Control Bulletin and the Meteorological Disaster Yearbook, which are from the website of the Ministry of Water Resources of China, and the global high-resolution atmospheric reanalysis dataset ERA5 (Fifth Generation ECMWF Global Climate Atmospheric Reanalysis) developed and maintained by the European Center for Medium-Range Weather Forecasts (ECMWF).

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

[0050] Specifically, due to the differences in satellite imaging in different regions and the limitations of extraction algorithms, many small water bodies cannot be captured. The spatiotemporal neighborhood similarity algorithm is used for further supplementation, and finally a complete half-month surface water product is obtained. For the desert snow and ice regions and mountain shadow regions existing in the initial half-month scale water body product, a mask processing is performed.

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

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

[0053]

[0054] wherein, is the total number of transitions of large water body to small water body, is the total number of transitions of non-water body to small water body.

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

[0056] (1) When long-time drought monitoring is performed, there are regional overcast, cloudy and rainy days, which leads to poor optical image interpretation integrity of DW water body product, and a large number of data missing areas appear, the random forest algorithm is used to extract water body data of Sentinel-1 SAR image, the missing areas of semi-monthly scale surface water are supplemented, and the initial semi-monthly scale water body product is obtained; the initial semi-monthly scale water body product is further supplemented by using the spatiotemporal neighborhood similarity algorithm, and the semi-monthly scale water body product is obtained.

[0057] Specifically, for the data missing area, the morphological filtering method is used to eliminate the scattering noise of the Sentinel-1 SAR image, five features of VV, VH, VV+VH, VV-VH and VV / VH are selected as sample features, 2000 water bodies and non-water bodies are manually selected as sample points, the random forest algorithm is used for classification, the supplementary water body of the Sentinel-1 SAR image is obtained, and the initial semi-monthly scale water body product is obtained by combining the water body data of DW.

[0058] Specifically, based on the algorithm of spatiotemporal neighborhood similarity, the water body of the area still having a gap is supplemented, according to the results of multiple experiments, the gap pixel with a radius of 70 m is selected as the spatial neighborhood, the similarity threshold is set to 30, the DW surface water data, the supplementary water body of the Sentinel-1 SAR image and the supplementary water body based on the spatiotemporal similarity are combined, and the final semi-monthly scale water body product is obtained.

[0059] Specifically, for the case of desert and ice and snow, the desert and ice and snow mask is used The specific implementation is as follows:

[0060]

[0061] 7

[0062]

[0063] wherein, is a normalized water index; is the reflectance value of the green band of the image; is the reflectance value of 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 the 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, precipitation gradually increases, and shallow water will form again. Adjacent small water bodies may merge into a complete large water body.

[0066] Based on the dynamic changes of 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 in pairs to form a water state transfer matrix. The definition is as follows:

[0067]

[0068]

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

[0070] Specifically, the extracted water products were converted into vector data, and each feature was divided into 10km grids, with the grid size greater 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 per square meter is assigned the value "SmallWater". Special processing is performed on non-water features. The water bodies in the 24 time periods of the year are merged to generate the maximum range of the year (MaxWater). The surface water and MaxWater of the year are then erased, and a category attribute is added with the value "NoWater". Ensure that the data for each time period contains three water categories, namely LargeWater, SmallWater, and NoWater.

[0071] Specifically, for each grid cell, the nearest features of the water body elements in the two adjacent periods are searched, the water body elements in the two periods are ensured to match accurately, and then the transition type attribute is added to the matched land features, for example, if the current state is small water body and the last state is large water body, it is represented as L_S.

[0072] Specifically, the six transition types in each grid are summed up one by one, and finally the water body state transition data of the 10 km spatial grid scale is obtained. Since the transition is calculated according to the water body products of the previous period and the next period, after the above calculation, a total of 23 transition matrices per year can be generated.

[0073] During the drought duration, large water bodies will gradually split into small water bodies, while during the drought relief, a large number of small water bodies will be formed and eventually merge into part of the large water body. Therefore, during the whole drought evolution process, the state transition related to small water bodies is particularly important, such as Figure 3 As shown. Through permutation and combination experiments of the six different transition types, and according to the above mechanism, two state conversion types with theoretically opposite trends, L_S and N_S, are selected to construct a new drought index.

[0074] Specifically, in order to highlight the difference between drought development and relief and to narrow the order of magnitude difference between different transition types, L_S and N_S are normalized by calculating the ratio of the difference to the sum of L_S and N_S; In addition, the smaller the drought index value, the higher the drought degree, and the normalized value is subtracted by 1 to finally obtain the normalized small water body state transition index NSWSTI.

[0075] Specifically, on the grid scale, only when the number of L_S and N_S in the grid is not zero, the index will be calculated, otherwise, it is considered as empty; On the basin scale, first, the sum of the number of L_S and N_S in each grid cell is accumulated, and then the index value of the whole study area is calculated according to the following formula:

[0076]

[0077]

[0078]

[0079]

[0080] wherein, and represent the number of transitions of L_S and N_S in grid cell i, respectively; and represent the total number of transitions of L_S and N_S in the basin, respectively; denotes the index value of a specific grid cell, and denotes the index value of the entire basin.

[0081] (4) Calculate the drought index of different time scales (n months), usually choose 1, 3, 6, 9 and 12 months, in order to match the NSWSTI results of the half-month scale water body product, calculate the SPI, SRI and SSMI of five scales, respectively denoted as 01, 03, 06, 09, 12, wherein 01 represents the half-month scale drought index, and 12 represents the half-year scale drought index.

[0082] Specifically, the Pearson correlation coefficient is a statistical index used to measure the degree of linear correlation between two variables X and Y, and its value is between -1 and 1, represented by R Value. The closer R is to 1 or -1, the higher the correlation between the two variables, and 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, it is considered that the linear relationship between the two variables is significant; if the p value is greater than 0.05, it is considered that the linear relationship between the two variables is not significant.

[0083] Specifically, after comparing the two basins, it is found that the correlation coefficient between the NSWSTI index and various drought indices in the Poyang Lake basin is higher than that in the Dongting Lake basin. Combined with the collected meteorological data, the SPI, SRI, SSMI drought indices are calculated, as shown in Table 1, the correlation between the NSWSTI index and SPI 01 is the highest, the Dongting Lake basin is 0.52, and the Poyang Lake basin is 0.54; 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 generally relatively low.

[0084] Specifically, the correlation coefficient between the NSWSTI index and the shallow soil moisture index (SSMI1 01) is relatively high, which is 0.46 in the Dongting Lake basin, but the correlation coefficient gradually decreases with the deepening of the soil layer, and it is only 0.24 in SSMI4 01. These results show that small shallow water bodies are highly sensitive to drought changes and are directly related to precipitation, and due to their limited ability to infiltrate deeper soil layers, they can effectively reflect precipitation patterns and describe drought conditions.

[0085] Specifically, from the perspective of drought indices of different time scales, the correlation of the NSWSTI index with the semi-monthly scale (SPI 01, SRI01 and SSMI1 01) is the highest, and the correlation decreases with the increase of the scale, and even cannot pass the significance test. Generally speaking, the drought index of a smaller time scale is more sensitive to meteorological drought, fully verifying the important role of small water bodies in short-term drought, and the NSWSTI index is more sensitive to the drought index of a short time scale, and is suitable for monitoring meteorological drought.

[0086] Table 1 Correlation coefficient of NSWSTI and traditional drought index

[0087]

[0088] The implementation basis of each embodiment of the present application is realized by the processing of the device with the processor function. Therefore, in engineering practice, the technical scheme and function of each embodiment of the present application are packaged into various modules. Based on this actual situation, on the basis of the above-mentioned embodiments, the embodiment of the present application provides a meteorological drought remote sensing monitoring system based on small water body transition matrix, which is used to execute the meteorological drought remote sensing monitoring method based on small water body transition matrix in the above-mentioned method embodiment.

[0089] The meteorological drought remote sensing monitoring system based on small water body transition matrix provided by the embodiment of the present application, as shown in Figure 4 The specific process of the embodiment includes the following steps: based on the obtained DW water body data, the semi-monthly scale surface water is extracted, the random forest algorithm and the spatio-temporal neighborhood similarity algorithm are used to supplement the extracted semi-monthly scale surface water in combination with the collected Sentinel-1 SAR image, the Sentinel-2 optical remote sensing image and the digital elevation model data, and the semi-monthly scale water body product is obtained; then the semi-monthly scale water body product is divided into three states based on the area size through the state conversion module, the six water body state conversion types are obtained through the spatial matching of the three states two by two, and the total number of transitions of the six water body state conversion types is respectively calculated; in the NSWSTI calculation module, the two types of large water body converted into small water body and non-water body converted into small water body which are theoretically opposite in change trend are selected, and the total number of transitions of the large water body converted into small water body and the non-water body converted into small water body is combined to construct the normalized small water body state conversion index NSWSTI; finally, through the drought monitoring module, the correlation analysis is performed in combination with the traditional drought index and the normalized small water body state conversion index NSWSTI, and the result shows that the drought index of a smaller time scale is more sensitive to meteorological drought, the NSWSTI index is more sensitive to the drought index of a short time scale, is suitable for remote sensing monitoring of meteorological drought, and more accurately reflects the occurrence, development and spatial expansion characteristics of drought, thereby providing important technical support for regional water resource management and ecological environment protection.

[0090] Based on the same inventive concept as the foregoing embodiments, the embodiments of the present application also provide an electronic device comprising a memory and a processor, the memory being configured to store computer executable instructions, and the processor being configured to execute the computer executable instructions to implement the small water body transfer matrix based remote sensing drought monitoring method as proposed in the foregoing embodiments.

[0091] The embodiments of the present application also provide a computer readable storage medium having a computer program stored thereon. The program, when executed by a processor, is configured to identify water body change signals in the early stage of drought more early, 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 a hard disk, a solid state disk, a flash disk, an optical disk or any non-volatile storage device, and is configured to store computer program codes and necessary data files. The stored computer program comprises a water body product acquisition module, a state transition module, an NSWSTI calculation module and a drought monitoring module.

[0092] Finally, it should be noted that the above specific embodiments are only representative examples of the present application. Apparently, the present application is not limited to the above specific embodiments, and can have many variations. Any simple modification, equivalent change and modification made to the above specific embodiments according to the technical essence of the present application shall be considered as falling within the protection scope of the present application.

Claims

1. A meteorological drought remote sensing monitoring method based on a small water body transfer matrix, characterized in that, The method comprises the following steps: Based on the obtained DW water body data, semi-monthly scale surface water is extracted, combined with the collected auxiliary water body data, and the extracted semi-monthly scale surface water is supplemented by using a random forest algorithm and a spatio-temporal neighborhood similarity algorithm to obtain a semi-monthly scale water body product; The semi-monthly scale water body product is divided into three states based on the area size, and six water body state conversion types are obtained by spatial matching of the three states, and the total number of transitions of the six water body state conversion types is calculated respectively to form a state transition matrix; The semi-monthly water body product is divided into three states based on the area size, including: water body area ≥ 10 5 square meters, recorded as large water body; water body area < 10 5 square meters and not connected to large rivers, recorded as small water body; other land use types except water body, recorded as non-water body; The two types of large water body converted into small water body and non-water body converted into small water body are selected, and the normalized small water body state conversion index NSWSTI is calculated based on the total number of transitions of the large water body converted into small water body and the non-water body converted into small water body, and the formula is: ， wherein, is the total number of transitions from large water bodies to small water bodies, is the total number of transitions from non-water bodies to small water bodies; Combined with the traditional drought index and the normalized small water body state conversion index NSWSTI, correlation analysis is performed, and meteorological drought remote sensing monitoring is performed based on the results of the correlation analysis.

2. The method according to claim 1, wherein, Combined with the collected auxiliary water body data, the extracted semi-monthly scale surface water is supplemented by using a random forest algorithm and a spatio-temporal neighborhood similarity algorithm, which comprises: The collected auxiliary water body data includes Sentinel-1 SAR image, Sentinel-2 optical remote sensing image, digital elevation model data, and maximum surface water range data; The random forest algorithm is used to extract the water body data of the Sentinel-1 SAR image, and the missing areas of the semi-monthly scale surface water are supplemented to obtain an initial semi-monthly scale water body product; The exponential threshold method is used for the Sentinel-2 optical remote sensing image to mask the desert and snow regions in the initial semi-monthly scale water body product; The spatio-temporal neighborhood similarity algorithm is used to further supplement the water body of the initial semi-monthly scale water body product; The mountain shadow layer data obtained from the digital elevation model data is used to mask the mountain shadow generated by the terrain height in the study area to obtain the final semi-monthly scale water body product.

3. The method according to claim 2, wherein, The exponential threshold method further includes the calculation formula is as follows: ， wherein, is the normalized water index, is the reflectance value of the green band of the image, is the reflectance value of the near infrared band.

4. The method according to claim 1, wherein, The semi-monthly scale water body product is divided into three states based on the area size, and the small water body is also masked by using the maximum surface water range data. The method comprises the following steps:

5. A remote sensing drought monitoring system based on small water body transfer matrix characterized in that, The water body product acquisition module is used to extract semi-monthly scale surface water based on the obtained DW water body data, combined with the collected auxiliary water body data, and the extracted semi-monthly scale surface water is supplemented by using a random forest algorithm and a spatio-temporal neighborhood similarity algorithm to obtain a semi-monthly scale water body product; The state conversion module is used to divide the semi-monthly scale water body product into three states based on the area size, and six water body state conversion types are obtained by spatial matching of the three states, and the total number of transitions of the six water body state conversion types is calculated respectively to form a state transition matrix; The NSWSTI calculation module is used to select the two types of large water body converted into small water body and non-water body converted into small water body, and the normalized small water body state conversion index NSWSTI is calculated based on the total number of transitions of the large water body converted into small water body and the non-water body converted into small water body, and the formula is: The semi-monthly water body product is divided into three states based on the area size, including: water body area ≥ 10 5 square meters, denoted as large water body; water body area < 10 5 square meters and not connected to large rivers and large rivers, denoted as small water body; other land use types other than water bodies, denoted as non-water body; ​ ， wherein, is the total number of transitions from large water bodies to small water bodies, is the total number of transitions from non-water bodies to small water bodies; 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 perform meteorological drought remote sensing monitoring based on the results of the correlation analysis. 6.An electronic device comprising a memory and a processor, the memory storing 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 according to any one of claims 1 to 4 are implemented.

7. A computer readable storage medium having stored thereon a computer program, 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 according to any one of claims 1 to 4 are implemented.

Images