Irrigation pond water storage monitoring method and device, electronic equipment and storage medium

Abstract

This invention provides a method, device, electronic equipment, and storage medium for monitoring the water storage capacity of irrigation ponds, belonging to the field of image processing technology. The method includes: determining the water surface area of ​​each irrigation pond within the target area based on initial remote sensing data of the target area, and acquiring the water level information of each irrigation pond; generating the water storage capacity of each irrigation pond based on its type, water surface area, and water level information. This invention utilizes object-oriented multi-scale segmentation combined with irrigation pond features for pond identification and water surface area calculation, while simultaneously collecting water depth data. This allows for rapid and timely calculation of accurate water storage capacity, improving monitoring efficiency and providing more intuitive water storage data. This provides fundamental data for irrigation district water volume accounting and optimized water allocation schemes, promoting economic benefits and water resource utilization efficiency.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, electronic device, and storage medium for monitoring the water storage capacity of irrigation pits and ponds. Background Technology

[0002] Monitoring the water storage capacity of irrigation ponds is an important part of assessing land drought resistance, and changes in water storage capacity are also an important reference indicator for water resource management.

[0003] Satellite remote sensing information has the characteristics of periodicity, macroscopicity and timeliness. Existing methods for monitoring the water storage of irrigation ponds usually use satellite remote sensing images to observe the water surface area and water storage of the ponds to establish a unified statistical relationship, and then estimate the water storage of each pond.

[0004] The above methods for monitoring ponds have drawbacks, including being unintuitive, having low accuracy, and having statistical relationships that are not suitable for some irrigation ponds. Summary of the Invention

[0005] The present invention provides a method, device, electronic equipment, and storage medium for monitoring the water storage capacity of irrigation pits and ponds. These methods address the shortcomings of existing pit and pond monitoring techniques, such as lack of intuitiveness, low accuracy, and incompatibility with certain irrigation pit and pond water bodies. The invention enables rapid and timely calculation of accurate water storage capacity in irrigation pits and ponds, improving monitoring efficiency while providing more intuitive water storage data. This data serves as a foundation for calculating irrigation water volume in irrigation districts and optimizing water allocation schemes, thereby promoting economic benefits and improving water resource utilization efficiency.

[0006] This invention provides a method for monitoring the water storage capacity of irrigation ponds, comprising:

[0007] Based on the initial remote sensing data of the target area, the water surface area of ​​each irrigation pond in the target area is determined, and the water level information of each irrigation pond is obtained; the water surface area is determined by multi-scale segmentation of the water surface of the irrigation ponds in the high-resolution remote sensing data in the initial remote sensing data;

[0008] Based on the irrigation pond type, water surface area, and water level information of each irrigation pond, the water storage capacity of each irrigation pond is generated. The irrigation pond types include regular and irregular types.

[0009] According to a method for monitoring water storage in irrigation ponds provided by the present invention, the initial remote sensing data includes multiple time-series low-resolution remote sensing data and single-temporal high-resolution remote sensing data; the resolution of the time-series low-resolution remote sensing data is lower than that of the high-resolution remote sensing data.

[0010] The determination of the water surface area of ​​each irrigation pond within the target area based on the initial remote sensing data of the target area includes:

[0011] In each time-series low-resolution remote sensing data, the irrigation pond area within the target area is extracted to generate the first irrigation pond dataset of all irrigation ponds within the target area.

[0012] Using the first irrigation pit dataset, the center point location of each irrigation pit within the target area is obtained;

[0013] The center point of each irrigation pit is migrated to the high-resolution remote sensing data;

[0014] In the high-resolution remote sensing data, a region mask is generated based on the center point of each irrigation pit as the center and a defined radius to obtain remote sensing data of the irrigation pit range; the defined radius is determined based on the maximum defined water surface area of ​​each irrigation pit in the first irrigation pit dataset;

[0015] Based on the segmentation scale threshold of irrigation pits and ponds, the irrigation pits and ponds in the remote sensing data of the irrigation pits and ponds range are merged into pixels to generate a second irrigation pit and pond dataset of all irrigation pits and ponds, and the water surface area of ​​the irrigation pits and ponds is obtained.

[0016] According to the present invention, a method for monitoring the water storage capacity of irrigation ponds is provided, which acquires water level information for each irrigation pond, including:

[0017] Using lidar, obtain water level information for irrigation ponds within the target area; or,

[0018] Acquire images of irrigation ponds within the target area to be identified;

[0019] The image to be identified is input into the water level recognition model to obtain the water level information output by the water level recognition model. The water level recognition model is obtained by training on sample pond images with water level information labels.

[0020] According to the present invention, a method for monitoring the water storage of irrigation ponds is provided, wherein irrigation pond areas are extracted from each time-series low-resolution remote sensing data, and a first irrigation pond dataset of all irrigation ponds in the target area is generated, including:

[0021] Using the information of large water bodies in the target area and the maximum NDVI obtained from the vegetation growth period of the time-series low-resolution remote sensing data, a mask is applied to any time-series low-resolution remote sensing data to generate masked remote sensing data of any time-series low-resolution remote sensing data, so as to obtain the masked remote sensing data of each time-series low-resolution remote sensing data; the masked remote sensing data includes remote sensing data of small water bodies and remote sensing data of soil.

[0022] The water body index is used to perform binary classification on each mask remote sensing data to obtain a small water body dataset within the target area in each mask remote sensing data.

[0023] The first irrigation pit dataset is determined based on the small water body dataset.

[0024] According to a method for monitoring the water storage capacity of irrigation ponds provided by the present invention, the first irrigation pond dataset is determined based on the small water body dataset, including:

[0025] Based on the pond size constraint, each water body in the small water body dataset of each scene of masked remote sensing data is filtered to obtain the filtered water body dataset for each time phase; the pond size constraint is determined based on the maximum defined water surface area of ​​the pond.

[0026] Spatial overlay analysis was performed on the selected water body datasets for all time periods to calculate the pixel-average water body frequency in the target region in order to determine the seasonal variation pattern of water volume in each small water body.

[0027] Based on the seasonal variation pattern of water volume, irregular irrigation pit and pond water body areas are determined; each irrigation pit and pond in the irregular irrigation pit and pond water body area is of irregular type.

[0028] Based on the shape characteristics of the irrigation pits and the irregular irrigation pit water body area, each water body in the filtered water body dataset is filtered to obtain the regular irrigation pit water body area; the irrigation pit type of each irrigation pit in the regular irrigation pit water body area is regular.

[0029] The combination of the irregular irrigation pond water body area and the regular irrigation pond water body area is used as the first irrigation pond dataset.

[0030] According to a method for monitoring the water storage capacity of irrigation ponds provided by the present invention, the water storage capacity of each irrigation pond is generated based on the pond type, water surface area, and water level information, including:

[0031] If any irrigation pond is of a regular type, the water storage capacity of any irrigation pond is determined based on the water level information and the water surface area of ​​any irrigation pond.

[0032] If any of the irrigation ponds is irregular in shape, the water storage capacity of any irrigation pond is determined based on the construction data of the irrigation pond, the water level information, or the water surface area of ​​the irrigation pond.

[0033] The present invention also provides a device for monitoring the water storage capacity of irrigation pits and ponds, comprising:

[0034] The determination module is used to determine the water surface area of ​​each irrigation pond within the target area based on the initial remote sensing data of the target area, and to obtain the water level information of each irrigation pond; the water surface area is determined based on multi-scale segmentation of the irrigation pond water surface of the high-resolution remote sensing data in the initial remote sensing data;

[0035] The generation module is used to generate the water storage capacity of each irrigation pit based on the type of irrigation pit, water surface area and water level information of each irrigation pit. The irrigation pit types include regular and irregular types.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the irrigation pond water storage monitoring method described above.

[0037] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the irrigation pond water storage monitoring method as described above.

[0038] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the irrigation pond water storage monitoring method as described above.

[0039] The present invention provides a method, device, electronic equipment, and storage medium for monitoring the water storage capacity of irrigation ponds. It utilizes object-oriented multi-scale segmentation combined with the characteristics of irrigation ponds to identify ponds and calculate the water surface area, while simultaneously collecting the water depth of the ponds. This allows for the rapid and timely calculation of the accurate water storage capacity of the ponds, improving monitoring efficiency and providing more intuitive water storage data. It is applicable to all irrigation ponds and provides basic data for water volume accounting in irrigation areas and optimization of water allocation schemes, thereby promoting economic benefits and improving water resource utilization efficiency. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0041] Figure 1 This is one of the flowcharts of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention;

[0042] Figure 2 This is a schematic diagram of the model training process provided by the present invention;

[0043] Figure 3 This is the second flowchart of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention;

[0044] Figure 4 This is the third flowchart of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention;

[0045] Figure 5 This is the fourth flowchart of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention;

[0046] Figure 6 This is a schematic diagram of the structure of the irrigation pit water storage monitoring device provided by the present invention;

[0047] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0049] Currently, most research on irrigation ponds focuses only on pond identification and water surface area extraction, with limited research on estimating pond water storage capacity. Furthermore, existing water storage estimation methods are largely based on the statistical relationship between area and water storage capacity. This method is ineffective for accurately estimating the water storage capacity of ponds where water level does not directly change with water surface area, especially when water level is unavailable. Accurately obtaining the water storage capacity of irrigation ponds requires three parts: pond identification, water surface area extraction, and water storage capacity calculation.

[0050] In the study of irrigation pond identification, the main methods include manual visual interpretation combined with the characteristics of irrigation ponds, automatic screening based on shape index, and object-oriented methods. Some studies utilize high-resolution NAIP data and Google imagery to visually interpret and digitize the location of irrigation ponds, identifying features such as the presence of irrigation regulating ditches, water supply pipes, drainage pipes, or pumping stations around the pond, as well as one or more adjacent farmlands, nearby ditches, and rivers—all sources of surface water. Other studies screen ponds based on their size and shape index characteristics, combined with OTU threshold segmentation for pond identification. Object-oriented methods for pond water body identification primarily rely on spectral feature bands and texture features, fused edge features, and semantic information for automatic pond identification.

[0051] In the extraction of pond surface area, many studies on detecting large-scale, long-term pond changes utilize optical imagery, such as the Theme Mapper (TM) / Enhanced Theme Mapper+ (ETM1) on Landsat, and the Moderate Resolution Imaging Spectroradiometer (MODIS) on the Earth Observing System (EOS) Terra and Aqua satellites. However, due to limitations such as low resolution and poor data quality, the extraction accuracy for small water bodies is relatively low. With the improvement of available satellite spatiotemporal resolution, more studies are using high spatiotemporal resolution imagery for precise water surface edge contour extraction, such as Sentinel-2, Planet, GF-1 / 2, Worldview, and Rapid Eye (RE). The emergence of high spatiotemporal resolution satellite data and SAR data, along with multi-source data fusion, provides more possibilities for the identification of small ponds. SAR data is unaffected by cloud cover and solar radiation, and can operate day and night under any weather conditions. Therefore, radar observation can serve as a substitute for or supplement to optical imagery in hydrological monitoring.

[0052] However, current research focuses primarily on extracting the surface area of ​​ponds and reservoirs, without incorporating water level and elevation information for water storage calculation. Some studies have added water level and elevation differences to the identification and extraction of pond water bodies. One approach involves creating a fusion classification model based on heterogeneous data from optical satellite images and lidar, using multi-level decision trees for identification. Alternatively, high-precision radar data can be used to acquire water levels. Regarding water storage calculation, multi-source remote sensing satellite data fusion has made significant contributions to addressing data gaps and improving accuracy. Most studies link multi-source elevation data with area estimation from optical satellites, establishing an empirical relationship between elevation and water surface area to calculate water storage, making multi-source data complementary in water storage estimation. Other studies combine satellite data with ground-based measured data for water storage calculation, fusing lidar or radar altimeter data to obtain water level information, establishing a relationship between water level and water surface area, and correcting and evaluating water levels using measured data. The calibrated water level and area information are then used to calculate water storage. However, the above method is not applicable to ponds where the water storage capacity does not change significantly with the water surface area, making it difficult to accurately estimate the water storage capacity of various types of irrigation ponds.

[0053] In particular, key water level information in water storage calculation is difficult to obtain. On the one hand, due to the small size of irrigation ponds, the method of obtaining water level using low-coverage lidar satellite remote sensing altimeters cannot be widely applied to all irrigation ponds, while radar satellite data with area coverage has the limitation of low resolution. On the other hand, the acquisition of data from actual hydrological stations is limited by the number and distribution of water level monitoring stations in different regions, and the large-scale setting up of monitoring instruments and control systems is time-consuming, labor-intensive, and costly, resulting in relatively limited current ground-based hydrological water level measurement data.

[0054] The above two factors limit the efficient monitoring of water storage in irrigation pits and ponds.

[0055] This invention discloses a method for estimating the water storage capacity of irrigation ponds by comprehensively considering remote sensing and ground observation. First, based on Sentinel-1 and Sentinel-2 data, OTU segmentation combined with irrigation pond features is used for rapid pond identification. Then, high-precision irrigation pond water surface area datasets are obtained through object-oriented multi-scale segmentation of high-resolution Gaofen-2 data from a single time phase. Next, water level information of irrigation ponds is obtained by combining ICESat2 lidar data and ground photographic observations. Finally, a method for estimating irrigation pond water storage capacity by fusing space and ground data is proposed. This invention proposes a multi-source data-supported method for estimating irrigation pond water storage capacity, providing fundamental data for irrigation district water volume accounting and exploring optimized water allocation schemes, thereby promoting improved economic benefits and water resource utilization efficiency.

[0056] This invention proposes a multi-source data-supported approach to estimate the water storage capacity of irrigation ponds, overcoming the drawback of difficulty in obtaining comprehensive water level information in the calculation of water storage capacity of irrigation ponds using a single method. It also solves the problem that the water storage capacity of irrigation ponds, which does not change with the water surface area, is difficult to calculate using the area-water storage statistical method.

[0057] Approach to estimating irrigation pond water storage using combined satellite and ground data

[0058] The following is combined Figures 1-7 The present invention describes the method, apparatus, electronic equipment, and storage medium for monitoring the water storage capacity of irrigation pits and ponds provided by embodiments of the present invention.

[0059] The irrigation pond water storage monitoring method provided in this embodiment of the invention can be implemented by an electronic device or software, functional module, or functional entity within an electronic device capable of implementing the irrigation pond water storage monitoring method. In this embodiment, the electronic device includes, but is not limited to, a server. It should be noted that the aforementioned implementation entity does not constitute a limitation of this invention.

[0060] Figure 1 This is one of the flowcharts illustrating the method for monitoring the water storage capacity of irrigation pits and ponds provided by this invention, such as... Figure 1 As shown, including but not limited to the following steps:

[0061] First, in step S1, based on the initial remote sensing data of the target area, the water surface area of ​​each irrigation pond in the target area is determined, and the water level information of each irrigation pond is obtained; the water surface area is determined by multi-scale segmentation of the water surface of the irrigation pond based on the high-resolution remote sensing data in the initial remote sensing data.

[0062] Two satellites, Sentinel 1 and Sentinel 2, were used to conduct long-term monitoring of the target area. Sentinel 1 collected two types of radar data, VV and VH polarization values, to determine whether it was an irrigation pond. Sentinel 2 collected optical remote sensing data in the near-infrared, red, and green light bands. The data from the two satellites corresponded to the SDWI and NDWI water indices, respectively.

[0063] The target area contains multiple irrigation ponds for irrigation. The initial remote sensing data is time-series remote sensing data collected by Sentinel 2 over a long period of time and stored in raster form. The data is then acquired by the GEE platform from Sentinel 2.

[0064] Based on the collected remote sensing data, each irrigation pond within the target area is identified, and the water surface area of ​​each irrigation pond is calculated using multi-scale segmentation. In addition, the water level information of each irrigation pond needs to be obtained. The water level information of each irrigation pond can be obtained by directly reading the water level of each irrigation pond, or by collecting it using lidar, or by identifying the water level in the collected images.

[0065] Optionally, the initial remote sensing data includes multiple time-series low-resolution remote sensing data and single-temporal high-resolution remote sensing data; the resolution of the time-series low-resolution remote sensing data is lower than that of the high-resolution remote sensing data.

[0066] The determination of the water surface area of ​​each irrigation pond within the target area based on the initial remote sensing data of the target area includes:

[0067] In each time-series low-resolution remote sensing data, the irrigation pond area within the target area is extracted to generate the first irrigation pond dataset of all irrigation ponds within the target area.

[0068] Using the first irrigation pit dataset, the center point location of each irrigation pit within the target area is obtained;

[0069] The center point of each irrigation pit is migrated to the high-resolution remote sensing data;

[0070] In the high-resolution remote sensing data, a region mask is generated based on the center point of each irrigation pit as the center and a defined radius to obtain remote sensing data of the irrigation pit range; the defined radius is determined based on the maximum defined water surface area of ​​each irrigation pit in the first irrigation pit dataset;

[0071] Based on the segmentation scale threshold of irrigation pits and ponds, the irrigation pits and ponds in the remote sensing data of the irrigation pits and ponds range are merged into pixels to generate a second irrigation pit and pond dataset of all irrigation pits and ponds, and the water surface area of ​​the irrigation pits and ponds is obtained.

[0072] The maximum defined water surface area can be determined by statistically analyzing a certain number of irrigation ponds and determining the largest area among all irrigation ponds. This is the limitation on the maximum area in the definition of an irrigation pond; for example, a maximum defined water surface area of ​​100,000 square meters can be set. First, in the initial remote sensing data, the irrigation pond areas in the target region are preliminarily extracted to obtain remote sensing data of the irrigation pond water bodies.

[0073] For example, the remote sensing data collected by Sentinel 1 and Sentinel 2 are time-series low-resolution remote sensing data with a resolution of 10 meters; the high-resolution remote sensing data collected by Gaofen 2 are single-temporal high-resolution remote sensing data, which includes multispectral remote sensing data and panchromatic data. The resolution of the multispectral remote sensing data is 4 meters, and the resolution of the panchromatic data is 1 meter.

[0074] The remote sensing data range of irrigation pits and ponds is larger than the second irrigation pit and pond dataset, which contains all irrigation pits and ponds. Through object-oriented multi-scale segmentation and merging, the water surface area of ​​the irrigation pits and ponds can be obtained.

[0075] Because the cost of acquiring high-resolution remote sensing data is too high and the processing speed is too slow, the irrigation pond water storage monitoring method provided by this invention first uses time-series low-resolution remote sensing data to delineate the area, and then uses single-temporal high-resolution remote sensing data to achieve accurate extraction.

[0076] Optionally, in each time-series low-resolution remote sensing data, the irrigation pond area within the target area is extracted to generate a first irrigation pond dataset containing all irrigation ponds within the target area, including:

[0077] Using the information of large water bodies in the target area and the maximum NDVI obtained from the vegetation growth period of the time-series low-resolution remote sensing data, a mask is applied to any time-series low-resolution remote sensing data to generate masked remote sensing data of any time-series low-resolution remote sensing data, so as to obtain the masked remote sensing data of each time-series low-resolution remote sensing data; the masked remote sensing data includes remote sensing data of small water bodies and remote sensing data of soil.

[0078] The water body index is used to perform binary classification on each mask remote sensing data to obtain a small water body dataset within the target area in each mask remote sensing data.

[0079] The first irrigation pit dataset is determined based on the small water body dataset.

[0080] The vegetation growing season can be the season when vegetation grows vigorously, such as June and July each year.

[0081] First, the area is masked to remove vegetation and large bodies of water.

[0082] Specifically, the Normalized Difference Vegetation Index (NDVI) of vegetation growth period is calculated using Sentinel 2 data, and a vegetation area mask for the current year is generated by limiting the threshold.

[0083] The Normalized Difference Vegetation Index (NDVI) of the target area during the vegetation growth period is obtained using Sentinel 2, and vegetation areas are determined by threshold limitation in order to remove vegetation areas.

[0084] The Normalized Difference Vegetation Index (NDVI) is calculated as follows:

[0085]

[0086] Where, ρ NIR Near-infrared reflectance; ρ RED This refers to the reflectivity in the infrared band.

[0087] After removing vegetation with NDVI and removing some large water areas with a mask, the remaining parts are bare soil and other water bodies, which can be well used for binary classification using the OTU thresholding method.

[0088] Then, based on existing geographic information of the target area such as GlobeLand30 and Open Street Map (OSM), masking is performed on large water bodies, wetlands, lakes, and irrigation canals to obtain masked area image files of small water bodies and bare soil areas.

[0089] Secondly, the OTU threshold segmentation method can be used to remove non-irrigated ponds and water bodies from remote sensing data, thereby achieving coarse extraction of water surface area.

[0090] The Otsu method (Maximum Inter-Class Difference) is an automatic thresholding method that selects the optimal threshold between water and bare soil. Applied to binary classification scenarios, the Otsu method is an automatic method based on pixel value distribution to distinguish between two relatively homogeneous objects. It is commonly used to differentiate between objects and background, bare soil and water, and forests and grasslands. The process involves calculating the proportions ω0 and ω1 of pixels with gray values ​​less than and greater than specific thresholds. The mean and variance of all pixels with gray values ​​less than the threshold are μ0, and the mean and variance of all pixels with gray values ​​greater than the threshold are μ1. Through iterative calculations, the threshold that maximizes the inter-class variance (BSS) is found. The specific calculation is as follows:

[0091] BSS=ω0ω1(μ0-μ1) 2 Formula 2

[0092] Within the aforementioned masked area, i.e., the area where irrigation ponds may appear, water bodies within the area, including irrigation ponds, tailrace ponds, and fishponds, are extracted using grayscale thresholding. The Saturated Difference Water Index (SDWI) is calculated using the VV and VH polarization values ​​from Sentinel 1 data, and the Normalized Difference Water Index (NDWI) is calculated using Sentinel 2 data. Then, the OTU thresholding method is used to identify the areas containing small artificial water bodies, obtaining a dataset of small water bodies within the target area. Simultaneously, the performance of the different indices in identifying irrigation ponds is compared, and the optimal thresholding index is selected.

[0093] SDWI = ln(10 × VV × VH) - 8 (Equation 3)

[0094]

[0095] Wherein, SDWI is the Sentinel-1 dual-polarized water index; VV is Sentinel-1 data used for vertical transmission and vertical reception (VV) dual-polarized data; VH is Sentinel-1 data used for vertical transmission and horizontal reception (VH) dual-polarized data; NDWI is the normalized water index; ρ GREEN Reflectance in the green band; ρ NIR This refers to the reflectivity in the near-infrared band.

[0096] Water quality indices SDWI and NDWI are used to achieve coarse extraction of irrigation pit and pond areas.

[0097] Optionally, the first irrigation pond dataset is determined based on the small water body dataset, including:

[0098] Based on the pond size constraint, each water body in the small water body dataset of each scene of masked remote sensing data is filtered to obtain the filtered water body dataset for each time phase; the pond size constraint is determined based on the maximum defined water surface area of ​​the pond.

[0099] Spatial overlay analysis was performed on the selected water body datasets for all time periods to calculate the pixel-average water body frequency in the target region in order to determine the seasonal variation pattern of water volume in each small water body.

[0100] Based on the seasonal variation pattern of water volume, irregular irrigation pit and pond water body areas are determined; the irrigation pit and pond type of each irrigation pit and pond in the irregular irrigation pit and pond water body area is irregular.

[0101] Based on the shape characteristics of the irrigation pits and the irregular irrigation pit water body area, each water body in the filtered water body dataset is filtered to obtain the regular irrigation pit water body area; the irrigation pit type of each irrigation pit in the regular irrigation pit water body area is regular.

[0102] The combination of the irregular irrigation pond water body area and the regular irrigation pond water body area is used as the first irrigation pond dataset.

[0103] Since the water volume of each irrigation pond varies at different times, it is necessary to calculate the pixel average water frequency to avoid omissions when statistically analyzing irrigation ponds using only remote sensing data.

[0104] For example, a maximum defined water surface area of ​​100,000 square meters can be set, and pond size restrictions can include water surface areas less than 100,000 square meters. Water body data with a surface area less than 100,000 square meters can be filtered out to form a filtered water body dataset for each time phase.

[0105] The first irrigation pit and pond dataset includes: irregular irrigation pit and pond water body regions and regular irrigation pit and pond water body regions, which are the results of coarse extraction of irrigation areas.

[0106] Irrigation pits and ponds were identified in the aforementioned water areas, taking into account factors such as the size and shape of the water surface and seasonal changes.

[0107] (1) For the area where small artificial water bodies are identified in the water body dataset, multiple small artificial water body layers extracted from February to June are used to perform spatial overlay analysis and calculate the average water body frequency of pixels in the area.

[0108] (2) Eliminate areas that are accidentally identified as water bodies, such as bare soil areas with high short-term humidity;

[0109] (3) Screening is carried out based on the characteristics of irrigation pits and ponds, including the size and shape of the pits and ponds.

[0110] Due to the relatively small area of ​​irrigation ponds, field survey data shows that the size of irrigation ponds in the region ranges from 5,000 square meters to 100,000 square meters. Based on this, some excessively large or small water bodies were eliminated; and the shape index S was limited. i To remove water bodies with extremely irregular shapes.

[0111] Finally, the seasonal water quality characteristics of irrigation ponds must also be considered.

[0112] For example, since the morphological characteristics of irrigation pits and ponds in the R irrigation area are mainly divided into two types: irregular water bodies and regular rectangular water bodies, the irregular water bodies are generally constructed of cement, with a small boundary slope and a gentle slope. The water surface area will change significantly with irrigation, precipitation and other processes. Therefore, irregular pits and ponds can be identified by recognizing the characteristics of water surface area changes, and water bodies with obvious early water storage-irrigation water use processes can be identified as irrigation pits and ponds.

[0113] However, some rectangular water bodies generally have steep boundary slopes and their surface areas do not change significantly. Therefore, after seasonal screening, the shape index S is used in the remaining areas. i As the external characteristics of irrigation pits and ponds, the type of irrigation pits and ponds is determined. If S i If S ≥ 0.5, then it is a regular type. i If <0.5, it is an irregular type.

[0114]

[0115] Where Si is the shape index; A is the area of ​​the pit; and P is the perimeter of the pit.

[0116] A dataset of irrigation ponds is generated based on Google Earth or other mapping software, as well as the results of field surveys. The determination of whether a particular data point is an irrigation pond and whether it is used for irrigation is based on two factors:

[0117] (1) Visually interpret the water area and determine whether it is an irrigation pit or pond based on the profile elevation;

[0118] (2) Determine whether there are relevant water facilities, such as wells, water rooms, and water outlets, on the high-resolution images of the map software.

[0119] Based on the above characteristics and field survey data, irrigation pits were identified, generating a ground verification dataset of 16 irrigation pits within the irrigation area. Comparing the extraction results based on SDWI and NDWI indices, the SDWI index threshold segmentation method based on Sentinel 1 and the NDWI index threshold segmentation method based on Sentinel 2 accurately identified irrigation pits within the region, achieving an overall accuracy of 94.1%.

[0120] Then, the semantic information of the object-oriented multi-scale segmentation algorithm is combined with water body indicators to accurately extract the surface area of ​​the pond.

[0121] Object-oriented multi-scale segmentation methods, combined with high-resolution remote sensing images, can accurately extract the boundary information of ground features. The basic principle is to divide pixels with the same features into an image object based on the shape, color, texture and other characteristics of the pixels.

[0122] Ecognition software employs object-oriented remote sensing analysis, utilizing multi-scale image segmentation to generate objects. It encapsulates spectral, shape, and texture characteristics for each object and establishes relationships between the object and its neighbors, parent objects, and child objects. Based on the generated information, it determines the type of ground cover and generates objects of different categories.

[0123] In multi-scale segmentation, the concept of scale refers to the granularity of the object to be processed from a macroscopic perspective, and can be understood from a microscopic perspective as the amount of heterogeneity allowed during the merging of objects in the current layer.

[0124] The hierarchy of image objects ranges from the pixel layer at the bottom to the entire scene of the image at the top. From bottom to top, the level of abstraction increases, and the heterogeneity (i.e., the segmentation scale) allowed for the objects increases.

[0125] Multi-scale image segmentation employs a region merging algorithm with minimal heterogeneity. It first merges pixels into smaller image objects, and then, through heterogeneity calculation and comparison, merges the smaller objects into larger objects layer by layer.

[0126] Each merge operation checks if the heterogeneity of the merged region is greater than the scale. If it is less, the merge occurs; otherwise, it does not. This process continues until the heterogeneity of the entire region is greater than the scale or all objects have been merged. In the EasiCom software, heterogeneity calculation primarily considers the spectral and shape characteristics within objects. The calculation formula is as follows:

[0127] f = ω color h color +(1-ω color )h shape Formula 6

[0128]

[0129] h sha =ω compact h compact +(1-ω compact )h smoo Formula 8

[0130]

[0131] h smooth =l / b Equation 10

[0132] Where f represents the total heterogeneity; ω color For spectral weights; (1-ω color ) represents the shape weight; ω compact For the firmness weight; (1-ω compact ) represents the smoothness weight; ω c For the spectral weights of the multispectral C-band; h color For spectral heterogeneity; h shape For shape heterogeneity; h compact For tightness; h smooth For smoothness; σ c denoted as , where is the standard deviation of the c-band spectral values ​​within object S; l is the number of pixels contained within the boundary of object S; n is the number of pixels contained within object S; and b is the boundary length of the minimum outer rectangle of object S.

[0133] In multi-scale segmentation, excessively large scales lead to incomplete segmentation, while excessively small scales result in overly fragmented segmentation. Therefore, it is necessary to determine the optimal scale threshold. The ESP2 plugin is used to find the peak value, and the corresponding threshold is the optimal scale threshold. For the identification of irrigation pits and ponds, the optimal scale threshold is 53.

[0134] Based on the obtained locations of irrigation ponds, and using an object-oriented approach, the high-resolution 2D dataset is used to accurately extract the water surface area of ​​the two types of irrigation ponds identified, generating a second irrigation pond dataset.

[0135] Because the irrigation ponds have small water areas, extracting the water surface area using Sentinel-2 remote sensing data with a resolution of 10 meters results in obvious jagged boundaries, causing significant area calculation errors. Therefore, based on Sentinel-2 data, only the irrigation ponds are identified and located, and the extracted center point of the identified irrigation ponds is spatially connected with the high-resolution image segmentation objects.

[0136] The Yikang software was used to perform multi-scale segmentation on the 1-meter resolution high-resolution 2 multispectral images after image fusion, and the NDWI index was combined for object classification, thereby achieving accurate extraction of water surface area and generating the second irrigation pond dataset.

[0137] Multi-scale segmentation algorithms based on high-resolution images are more accurate than sentinel extraction, and their edge processing is more detailed, resulting in a more precise water surface area. However, segmentation based on the shape, color, and texture of objects has higher requirements for image quality. If the image has high cloud content or poor image quality, it can lead to errors in the boundary extraction results.

[0138] The accuracy of the water surface area extraction method for irrigation ponds is evaluated by comparing the water surface area obtained by remote sensing with the water surface area estimated by actual water level.

[0139] For irrigation pits with flat bottoms and regular shapes, an irregular triangular network (TIN) model of the irrigation pit can be constructed using known construction parameters. For any input water level, its water surface area and water storage capacity can be calculated.

[0140] The average water level measured in the field over multiple periods can be used as the true ground value of the water level for the immediate period. Using the actual water level as input data, and combining it with the TIN model, the true water surface area for the current period is obtained. This value is then compared with the water surface area estimated by remote sensing to evaluate the water surface area extraction method. The method was validated on the water surface areas of five irrigation ponds, with RMSE and MAE of 962 square meters and 766 square meters, respectively. The extraction error was within ±1060 square meters. Considering the resolution of the Gaofen-2 image after image fusion (1 meter), the overall extraction error was within the allowable range of resolution error. Therefore, the method of extracting the water surface area of ​​irrigation ponds by combining Sentinel-2 and high-resolution remote sensing imagery is feasible.

[0141] Optionally, obtaining the water level information of each irrigation pond includes:

[0142] Using lidar, obtain water level information for irrigation ponds within the target area; or,

[0143] Acquire images of irrigation ponds within the target area to be identified;

[0144] The image to be identified is input into the water level recognition model to obtain the water level information output by the water level recognition model. The water level recognition model is obtained by training on sample pond images with water level information labels.

[0145] Water level information is obtained using lidar data.

[0146] IceSat2 lidar data can provide ground elevations (relative to the geoid) for land, water, and vegetation. The acquired nearby surface elevation is used to replace the upper boundary elevation of the irrigation pit, and the difference between this value and the water surface is calculated. This difference, combined with the construction depth of the irrigation pit, is then used to calculate the current water depth. The elevation H of the boundary around the irrigation pit is also used. b and water surface elevation H w The difference between the two values ​​yields the current water surface height h1 from the upper boundary of the construction site. Then, the obtained construction depth h of the irrigation pit is used to calculate the height. c The water depth h can be obtained by subtracting h1 from h1. w The details are as follows:

[0147] H b -H w =h1 Equation 11

[0148] h c -h1=h w Formula 12

[0149] Among them, H b and H w These are the elevations of the surrounding boundary of the irrigation pit and the water surface; h1 is the current height of the water surface from the upper boundary of the structure; h cIt refers to the depth of the irrigation pits and ponds; h w It's because the water is deep.

[0150] To obtain water levels through image observation, the acquisition of the photo set first requires certain constraints to ensure the comparability of the photos. This means ensuring that the series of photos in the images to be identified are taken from the same camera, have a fixed viewpoint, a fixed distance, are from adjacent dates, and have consistent weather conditions.

[0151] Figure 2 This is a schematic diagram of the model training process provided by the present invention, as shown below. Figure 2 As shown, firstly, a part needs to be consistent with the original. Figure 1 ,Original Figure 2 The original figure n corresponds to the measured water level data.

[0152] Then, the Otsu thresholding method was used to segment the original... Figure 1 ,Original Figure 2 The original image n is preprocessed to... Figure 1 ,Original Figure 2 ... Convert the original image n to grayscale Figure 1 Grayscale Figure 2 ..., grayscale image n, then the grayscale... Figure 1 Grayscale Figure 2 ..., parallel smoothing of grayscale image n yields a smooth image. Figure 1 ,smooth Figure 2 Smooth graph n, ..., smooth graph n.

[0153] Next, water level recognition features are acquired, mainly including the lake surface area and the leftmost / rightmost edge pixel column where the lake surface is photographed. By setting the center point of the lake surface and combining breadth-first search of the lake surface area and the rightmost or leftmost edge pixel column of the lake surface (determined according to the shooting angle of the photo), the input feature parameters Feature 1 and Feature 2 for model training are obtained.

[0154] Finally, a water level identification model based on back-propagation neural networks (BPNNs) was used for water level prediction. During the model training phase, 80% of the sample set was used to train the image features, and 20% was used for validation.

[0155] The water level identification model includes an input layer, a hidden layer, and an output layer. The input quantity of the input layer is Xj, the hidden layer output quantity of the hidden layer is Oj, and the output layer output quantity is Yk.

[0156] The accuracy of two water level acquisition methods was verified using measured water level data. The measured ground water depth was compared with water levels obtained from lidar data and automatically acquired from photographs to analyze the accuracy of water depth measurements based on remote sensing and ground observation methods. Irrigation pond No. 1, which had data from ICESat2 orbital data, ground construction data, and measured ground data, was selected for verification of the water level acquisition methods. The average of the April field water level measurements was taken as the true ground water level value. The water level obtained from ICESat2 data was 1.04 meters, while the water level observed from photographs was 1.10 meters. The water level obtained from ICESat2 was more accurate than the water level observed from photographs, with an error of 2 centimeters, compared to 8 centimeters for photographic observation.

[0157] The construction data includes: degree of inclination, bottom area, opening area, area and depth of slope variation layers, etc.

[0158] Further, in step S2, the water storage capacity of each irrigation pit is generated based on the irrigation pit type, water surface area and water level information of each irrigation pit. The irrigation pit type includes regular type and irregular type.

[0159] Among them, the surface area of ​​regular irrigation ponds does not change with the water storage volume, while the surface area of ​​irregular irrigation ponds changes with the water storage volume.

[0160] For regular-shaped irrigation ponds, the water storage capacity can be directly obtained by multiplying the water surface area and water level information. For all irrigation ponds, the water storage capacity can be obtained by using the water surface area and the TIN model of the bottom morphology of the irrigation pond generated by combining the ground data with the water surface area, or by using the water level information and the TIN model of the bottom morphology of the irrigation pond generated by combining the ground data with the water level of the irrigation pond.

[0161] Optionally, the water storage capacity of each irrigation pond is generated based on its pond type, water surface area, and water level information, including:

[0162] If any irrigation pond is of a regular type, the water storage capacity of any irrigation pond is determined based on the water level information and the water surface area of ​​any irrigation pond.

[0163] If any of the irrigation ponds is irregular in shape, the water storage capacity of any irrigation pond is determined based on the construction data of the irrigation pond, the water level information, or the water surface area of ​​the irrigation pond.

[0164] The combination of image recognition and radar measurement, along with data from the construction of irrigation ponds, allows for the direct calculation of water storage capacity in irrigation ponds. This method is applicable to both types of irrigation ponds, especially those where water storage capacity does not change with the surface area.

[0165] Furthermore, for irrigation ponds where water level information cannot be directly obtained and water storage varies with the water surface area, precise water surface area identification data is used, combined with a TIN model of the pond bottom morphology generated from ground data, to calculate the water storage based on the water surface area-water storage relationship. The selection of these two different methods for obtaining irrigation pond water levels depends on data availability, processing efficiency, and final accuracy. Using a combination of these three water storage calculation approaches can ensure accurate water storage calculations for various types of irrigation ponds under different conditions.

[0166] (1) Calculate the water level depth based on lidar data and irrigation pit construction data, and then calculate the water storage capacity based on the precisely extracted water surface area information of the irrigation pit.

[0167] (2) The water level of irrigation pits and ponds is automatically obtained through a series of water surface change photos, and the water storage capacity is calculated by combining the accurately extracted water surface area information of irrigation pits and ponds.

[0168] (3) Utilize the relationship between the construction data of irrigation pits and ponds and the changes in water surface area, and combine it with the current water surface area to obtain the water storage capacity.

[0169] For irrigation ponds whose surface area changes with the water level, all three methods can be used to calculate the water storage capacity; for irrigation ponds whose surface area does not change with the water level, only the first two methods can be used to obtain the water level and calculate the water storage capacity.

[0170] Each method of obtaining water level has its advantages and disadvantages:

[0171] The method of extracting water level information based on the LiDAR ICESat2 is faster, but because the water body of the irrigation pit is small, there will be cases where no trajectory passes through during the key observation time interval, resulting in missing data and making it impossible to calculate the water storage of the irrigation pit.

[0172] While photo-based observation methods require some preliminary sample collection and instrument setup, they can ensure timely and efficient acquisition of water level information for all irrigation ponds. By automatically identifying multiple photos and combining them with accurately extracted water surface area, they can supplement and guarantee water storage calculations in cases where remote sensing data is missing. Photo-based observation methods are preferred for regular-shaped irrigation ponds.

[0173] (1) For the estimation of water storage in regular irrigation ponds, it is necessary to obtain the water surface area and the current water level separately to calculate the water storage. Therefore, the water level can only be obtained by observing LiDAR data and water surface photos, and then the water storage can be calculated by combining the accurately extracted water surface area information of the irrigation pond.

[0174] (2) For calculating the water storage capacity of irregular irrigation ponds, as long as the current water level or the current water surface area is obtained, the water storage capacity of the irrigation pond can be calculated by combining the TIN model of the bottom morphology of the irrigation pond generated by the ground data, based on the water level-water storage capacity and area-water storage capacity lookup table relationships, respectively. Therefore, all three methods mentioned above can be used to calculate the water storage capacity of irrigation ponds. The area-water storage capacity relationship method has the advantage of large-scale area recognition, and the area recognition error has a smaller impact on the result than the water level recognition error, resulting in higher accuracy. For regular irrigation ponds, the area-water storage capacity relationship method should be used first.

[0175] The accuracy of water storage calculation results for two different types of irrigation ponds was evaluated. For two measured irrigation ponds, the water storage calculated based on the method was compared with the results calculated based on ground measurements. Irrigation pond 1 represents the type where the water surface area does not change with the water level, while irrigation pond 2 represents the type where the water surface area changes with the water level. The results show that the accuracies of the area-water storage calculation method, the ICESat2 (LiDAR)-water storage calculation method, and the photo observation-water storage calculation method are 98.8%, 95.2%, and 94.1%, respectively, all demonstrating good performance.

[0176] The irrigation pond water storage monitoring method provided by this invention utilizes object-oriented multi-scale segmentation combined with irrigation pond characteristics to identify irrigation ponds and calculate water surface area, while simultaneously collecting water depth data. This allows for rapid and timely calculation of accurate irrigation pond water storage, improving monitoring efficiency and providing more intuitive water storage data. This provides fundamental data for irrigation district water volume accounting and optimizing water allocation schemes, thereby promoting economic benefits and water resource utilization efficiency.

[0177] Figure 3 This is the second flowchart illustrating the method for monitoring the water storage capacity of irrigation pits and ponds provided by this invention. Figure 3 As shown, it includes:

[0178] Irrigation ponds were identified using data collected by Sentinel 1 and Sentinel 2.

[0179] The dual-polarized water index SDWI and normalized water index NDWI were constructed respectively for water body extraction;

[0180] Furthermore, based on the area and shape characteristics of irrigation pits and ponds and the seasonal changes in water surface area, the irrigation pits and ponds can be extracted;

[0181] Based on the Gaofen-2 data, the water surface area of ​​the irrigation ponds was accurately extracted. Specifically, an object-oriented multi-scale segmentation method was used to obtain the water surface area of ​​the irrigation ponds.

[0182] For regular irrigation pits, IceSat2 is used to build data or observe photos to obtain water level information of the irrigation pits, then the water storage capacity of the irrigation pits is estimated, and finally the water storage capacity of the irrigation pits is obtained.

[0183] For irregularly shaped irrigation pits, the water storage capacity of the irrigation pits is estimated, and the final water storage capacity of the irrigation pits is obtained.

[0184] Figure 4 This is the third flowchart of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention, as shown below. Figure 4 As shown, the process includes: monitoring the water level of irrigation pits and ponds using lidar and photographic observation. Specifically, the lidar data is preprocessed to obtain elevation points, which in turn determine the height from the water surface to the top of the pit or pond. The water depth is then calculated by combining this with the construction depth. Samples are collected and preprocessed, and water level features are extracted. These water level features are then input into a backpropagation neural network for water level prediction.

[0185] For irregular irrigation pits, a TIN model of the bottom of the irrigation pit is constructed, and the dynamic water surface area is obtained to obtain the water depth. Finally, the water storage capacity is calculated.

[0186] For regular-shaped irrigation ponds, the water surface area is obtained first, then the water depth is obtained, and finally the water storage capacity is calculated.

[0187] Figure 5 This is the fourth flowchart of the method for monitoring the water storage capacity of irrigation pits and ponds provided by the present invention, as shown below. Figure 5 As shown, it includes:

[0188] First, in step 501, irrigation pits and ponds are identified;

[0189] Secondly, in step 502, the surface area of ​​the irrigation pit is accurately extracted;

[0190] Furthermore, in step 503, the irrigation pond water level information is obtained through the fusion of space and ground data;

[0191] Finally, in step 504, the water storage capacity of irrigation ponds is estimated by multi-source and data fusion.

[0192] This invention proposes three methods for calculating the water storage capacity of irrigation ponds by constructing a remote sensing model and combining satellite data and ground photographic observations. These methods integrate water surface area, lidar water level acquisition, and water level acquisition through photographic observation to effectively calculate the water storage capacity of both regular and irregular-shaped irrigation ponds. In particular, the method based on ground photographic observations solves the problem that the single method of water surface area minus water storage capacity is insufficient for calculating the water storage capacity of irrigation ponds, overcoming the limitations of single models and single data sources, and effectively improving the accuracy and coverage of irrigation pond water storage capacity calculations.

[0193] The irrigation pond water storage monitoring device provided by the present invention is described below. The irrigation pond water storage monitoring device described below can be referred to in correspondence with the irrigation pond water storage monitoring method described above.

[0194] Figure 6 This is a schematic diagram of the structure of the irrigation pond water storage monitoring device provided by the present invention, as shown below. Figure 6 As shown, it includes:

[0195] The determination module 601 is used to determine the water surface area of ​​each irrigation pond in the target area based on the initial remote sensing data of the target area, and to obtain the water level information of each irrigation pond; the water surface area is determined based on the object-oriented multi-scale segmentation of the water surface of the irrigation pond in the high-resolution remote sensing data in the initial remote sensing data;

[0196] The generation module 602 is used to generate the water storage capacity of each irrigation pit based on the irrigation pit type, water surface area and water level information of each irrigation pit. The irrigation pit type includes regular type and irregular type.

[0197] During the operation of the device, the determination module 601 determines the water surface area of ​​each irrigation pond within the target area based on the initial remote sensing data of the target area, and obtains the water level information of each irrigation pond; the water surface area is determined based on object-oriented multi-scale segmentation of the water surface of the irrigation ponds in the high-resolution remote sensing data in the initial remote sensing data; the generation module 602 generates the water storage capacity of each irrigation pond according to the irrigation pond type, water surface area and water level information of each irrigation pond, wherein the irrigation pond type includes regular and irregular types.

[0198] The irrigation pond water storage monitoring device provided by this invention utilizes object-oriented multi-scale segmentation combined with the characteristics of irrigation ponds to identify irrigation ponds and calculate the water surface area. At the same time, it collects the water depth of irrigation ponds, thereby quickly and timely calculating the accurate water storage of irrigation ponds. This improves monitoring efficiency and provides more intuitive water storage data, providing basic data for irrigation area water volume accounting and optimization of water allocation schemes, and promoting the improvement of economic benefits and water resource utilization efficiency.

[0199] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 7 As shown, the electronic device may include a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute a method for monitoring the water storage of irrigation ponds. This method includes: determining the water surface area of ​​each irrigation pond in the target area based on initial remote sensing data of the target area, and acquiring the water level information of each irrigation pond; the water surface area is determined based on object-oriented multi-scale segmentation of the water surface of the irrigation ponds in the high-resolution remote sensing data of the initial remote sensing data; and generating the water storage capacity of each irrigation pond according to the type of irrigation pond, water surface area, and water level information of each irrigation pond, wherein the irrigation pond type includes regular and irregular types.

[0200] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0201] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the irrigation pond water storage monitoring method provided by the above methods. The method includes: determining the water surface area of ​​each irrigation pond in the target area based on initial remote sensing data of the target area, and obtaining the water level information of each irrigation pond; the water surface area is determined based on multi-scale segmentation of the water surface of the irrigation ponds in the high-resolution remote sensing data in the initial remote sensing data; and generating the water storage capacity of each irrigation pond according to the irrigation pond type, water surface area, and water level information of each irrigation pond, wherein the irrigation pond type includes regular and irregular types.

[0202] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the irrigation pond water storage monitoring method provided by the above methods. The method includes: determining the water surface area of ​​each irrigation pond in the target area based on initial remote sensing data of the target area, and acquiring the water level information of each irrigation pond; the water surface area is determined based on multi-scale segmentation of the water surface of the irrigation ponds in the high-resolution remote sensing data in the initial remote sensing data; and generating the water storage capacity of each irrigation pond according to the irrigation pond type, water surface area, and water level information of each irrigation pond, wherein the irrigation pond type includes regular and irregular types.

[0203] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0204] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0205] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of monitoring the water volume of an irrigation pond, characterized in that, The method comprises the following steps: Based on the initial remote sensing data of the target area, the water surface area of each irrigation pond in the target area is determined, and the water level information of each irrigation pond is obtained; the water surface area is determined based on the multi-scale segmentation of the water surface of the irrigation pond in the high-resolution remote sensing data in the initial remote sensing data; According to the irrigation pond type, water surface area and water level information of each irrigation pond, the water storage capacity of each irrigation pond is generated, and the irrigation pond type includes regular type and irregular type; The initial remote sensing data includes a plurality of time-series low-resolution remote sensing data and a single-time high-resolution remote sensing data; the resolution of the time-series low-resolution remote sensing data is lower than that of the high-resolution remote sensing data; Based on the initial remote sensing data of the target area, the water surface area of each irrigation pond in the target area is determined, which comprises: In each time-series low-resolution remote sensing data, the irrigation pond area in the target area is extracted to generate a first irrigation pond data set of all irrigation ponds in the target area; Using the first irrigation pond data set, the center point position of each irrigation pond in the target area is obtained; The center point position of each irrigation pond is migrated to the high-resolution remote sensing data; In the high-resolution remote sensing data, the center point position of each irrigation pond is taken as the center, and a region mask is generated according to the limited radius to obtain the irrigation pond range remote sensing data; the limited radius is determined based on the maximum defined water surface area of each irrigation pond in the first irrigation pond data set; Based on the segmentation scale threshold of the irrigation pond, the irrigation pond in the irrigation pond range remote sensing data is merged to generate a second irrigation pond data set of all irrigation ponds, and the water surface area of the irrigation pond is obtained; In each time-series low-resolution remote sensing data, the irrigation pond area in the target area is extracted to generate a first irrigation pond data set of all irrigation ponds in the target area, which comprises: Using the large water body information of the target area and the maximum NDVI obtained in the vegetation growth period from the time-series low-resolution remote sensing data, a mask is generated for any time-series low-resolution remote sensing data to generate mask remote sensing data of the any time-series low-resolution remote sensing data, so as to obtain mask remote sensing data of each time-series low-resolution remote sensing data; the mask remote sensing data includes small water body remote sensing data and soil remote sensing data; Using the water body index, each scene mask remote sensing data is binarized to obtain a small water body data set in the target area in each scene mask remote sensing data; According to the small water body data set, the first irrigation pond data set is determined.

2. The method of claim 1, wherein, The water level information of each irrigation pond is obtained, which comprises: Using laser radar, the water level information of the irrigation pond in the target area is obtained; or Obtaining the to-be-recognized image of the irrigation pond in the target area; The to-be-recognized image is input into a water level recognition model to obtain the water level information output by the water level recognition model, and the water level recognition model is obtained based on the sample pond image with water level information label.

3. The method of claim 1, wherein, According to the small water body data set, the first irrigation pit pond data set is determined, including: According to the pit pond size limit condition, each water body in the small water body data set of each scene mask remote sensing data is screened, and a screened water body data set of each time phase is obtained; the pit pond size limit condition is determined based on the maximum defined water surface area of the pit pond water body; Spatial overlay analysis is performed on the screened water body data sets of all time phases, and pixel average water body frequency in the target area is calculated to determine the water quantity seasonal variation law of each small water body; According to the water quantity seasonal variation law, an irregular type irrigation pit pond water body region is determined; the irrigation pit pond type of each irrigation pit pond in the irregular type irrigation pit pond water body region is irregular type; According to the irrigation pit pond contour feature and the irregular type irrigation pit pond water body region, each water body in the screened water body data set is screened to obtain the regular type irrigation pit pond water body region; the irrigation pit pond type of each irrigation pit pond in the regular type irrigation pit pond water body region is regular type; The combination of the irregular type irrigation pit pond water body region and the regular type irrigation pit pond water body region is taken as the first irrigation pit pond data set.

4. The method of claim 1-3, wherein, According to the pit pond type, water surface area and water level information of each irrigation pit pond, the water storage capacity of each irrigation pit pond is generated, including: In the case that the type of any irrigation pit pond is regular type, the water storage capacity of the any irrigation pit pond is determined according to the water level information and the water surface area of the any irrigation pit pond; In the case that the type of the any irrigation pit pond is irregular type, the water storage capacity of the any irrigation pit pond is determined according to the water level information or the water surface area of the any irrigation pit pond based on the construction data of the any irrigation pit pond.

5. An apparatus for monitoring the water storage capacity of an irrigation pond, characterized by It includes: A determination module is configured to determine water surface areas of each irrigation pit pond in a target area based on initial remote sensing data of the target area, and obtain water level information of each irrigation pit pond; the water surface area is determined based on multi-scale segmentation of irrigation pit pond water surfaces in high-resolution remote sensing data in the initial remote sensing data; A generation module is configured to generate water storage capacities of each irrigation pit pond according to pit pond types, water surface areas and water level information of each irrigation pit pond, the pit pond type including regular type and irregular type; The initial remote sensing data includes a plurality of time-series low-resolution remote sensing data and a single-time-phase high-resolution remote sensing data; the resolution of the time-series low-resolution remote sensing data is lower than that of the high-resolution remote sensing data; The determination of the water surface areas of each irrigation pit pond in the target area based on the initial remote sensing data of the target area includes: In each time-series low-resolution remote sensing data, an irrigation pit pond region in the target area is extracted to generate a first irrigation pit pond data set of all irrigation pit ponds in the target area; The center point position of each irrigation pit pond is obtained by using the first irrigation pit pond data set; The center point position of each irrigation pit pond is migrated to the high-resolution remote sensing data; In the high-resolution remote sensing data, a region mask is generated based on a center point position of each irrigation pond and a defined radius, and the irrigation pond range remote sensing data is acquired; the defined radius is determined based on a maximum defined water surface area of each irrigation pond in the first irrigation pond data set; Based on a segmentation scale threshold of the irrigation pond, the irrigation ponds in the irrigation pond range remote sensing data are merged by pixel, and a second irrigation pond data set of all the irrigation ponds is generated, and the water surface area of the irrigation pond is acquired; In each time-series low-resolution remote sensing data, an irrigation pond region in the target region is extracted, and a first irrigation pond data set of all the irrigation ponds in the target region is generated, including: Using the largest NDVI in the vegetation growth period acquired by the large water body information of the target region and the time-series low-resolution remote sensing data, a mask is generated for any time-series low-resolution remote sensing data, and mask remote sensing data of the any time-series low-resolution remote sensing data is generated, so as to acquire mask remote sensing data of each time-series low-resolution remote sensing data; the mask remote sensing data includes small water body remote sensing data and soil remote sensing data; Using the water body index, each scene mask remote sensing data is binarized, and a small water body data set in the target region in each scene mask remote sensing data is acquired; According to the small water body data set, the first irrigation pond data set is determined.

6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the irrigation pond water storage monitoring method of any one of claims 1-4.

7. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the irrigation pond water storage monitoring method of any one of claims 1-4.

Images