Riverbank Collapse Detection Method Based on Remote Sensing Inversion Time Series Data

Abstract

This application relates to a method for detecting riverbank collapse based on remote sensing inversion time-series data. The method includes: acquiring remote sensing inversion time-series data of a target riverbank formed within a preset detection period, the remote sensing inversion time-series data including time-series data of suspended sediment concentration in multiple pixel areas covered by the target riverbank; identifying potential collapse areas from multiple pixel areas based on all suspended sediment concentration time-series data; selecting qualified remote sensing inversion data for potential collapse areas from the target suspended sediment concentration time-series data based on the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data; determining the data quality index of suspended sediment concentration in potential collapse areas at each detection time-series based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of potential collapse areas within the preset detection period; and performing collapse detection on the target riverbank based on the data quality index. This method improves the accuracy of riverbank collapse detection.

Description

Technical Field

[0001] This application relates to the field of computer vision technology, and in particular to a method for detecting riverbank collapse based on remote sensing inversion time series data. Background Technology

[0002] Riverbank collapse is a common phenomenon in river geomorphological evolution. Once a riverbank collapses, it poses a serious threat to river stability, flood control safety, and riverside infrastructure. It not only directly causes land loss, but also leads to frequent changes in river morphology or waterway blockage, and may even trigger more severe flood disasters. Therefore, it is essential to conduct riverbank collapse detection.

[0003] Currently, riverbank collapse detection typically relies on qualitative identification of riverbank topography, employing methods such as manual inspection and visual recording, image difference analysis, visual interpretation, terrain-based identification, and model simulation. However, because riverbank topography is susceptible to environmental interference and spatiotemporal changes, collapse detection often fails to reflect the actual collapse situation, leading to false alarms. Consequently, the accuracy of current riverbank collapse detection methods is low. Summary of the Invention

[0004] Therefore, it is necessary to provide a riverbank collapse detection method based on remote sensing inversion time series data to improve the accuracy of riverbank collapse detection and address the aforementioned technical problems.

[0005] Firstly, this application provides a method for detecting riverbank collapse based on remote sensing inversion time-series data, including:

[0006] The remote sensing inversion time series data of the target riverbank formed within a preset detection period is obtained, wherein the remote sensing inversion time series data includes the time series data of suspended sediment concentration corresponding to each of the multiple pixel areas covered by the target riverbank.

[0007] Based on all time-series data of suspended sediment concentration, potential collapse areas are identified from the multiple pixel regions.

[0008] Based on the spatiotemporal variation characteristics of the target suspended sediment concentration time series data of the potential collapse area, qualified remote sensing inversion data of the potential collapse area are selected from the target suspended sediment concentration time series data.

[0009] Based on the correlation between the total remote sensing observation data of the potential collapse area during the preset detection period and the qualified remote sensing inversion data, the data quality index of the suspended sediment concentration of the potential collapse area at each detection time is determined.

[0010] Based on the data quality indicators, the target riverbank is subjected to collapse detection.

[0011] In one embodiment, identifying potential landslide areas from the plurality of pixel regions based on all time-series data of suspended sediment concentration includes:

[0012] Extract multiple suspended sediment concentration values ​​corresponding to each pixel region from all the time-series data of suspended sediment concentration.

[0013] Based on all the time-series data of suspended sediment concentration, determine the suspended sediment concentration threshold corresponding to each of the multiple pixel regions.

[0014] Based on the relationship between the concentration values ​​of all suspended sediments in each pixel region and the suspended sediment concentration threshold, potential collapse areas are identified from the multiple pixel regions.

[0015] In one embodiment, determining the suspended sediment concentration threshold corresponding to each of the plurality of pixel regions based on the time-series data of all suspended sediment concentrations includes:

[0016] The time-series data of suspended sediment concentration in each pixel region are sorted according to the magnitude of the suspended sediment concentration value to obtain the target suspended sediment concentration time-series data arranged sequentially in each pixel region.

[0017] Based on the effective data volume and preset percentile of the time series data of the target suspended sediment concentration, determine the target position of the preset percentile in the time series data of the target suspended sediment concentration.

[0018] The target suspended sediment concentration value corresponding to the target location is located in the time series data of the target suspended sediment concentration, and the target suspended sediment concentration value is used as the suspended sediment concentration threshold of each pixel region.

[0019] In one embodiment, the spatiotemporal variation characteristics include attribute status identifiers; the step of selecting qualified remote sensing inversion data of the potential collapse area from the time series data of the target suspended sediment concentration based on the spatiotemporal variation characteristics of the target suspended sediment concentration in the potential collapse area includes:

[0020] Based on the attribute status identifiers, generate an attribute status identifier sequence for the target suspended sediment concentration time series data at all detection times of suspended sediment concentration.

[0021] From the time series data of the target suspended sediment concentration, the time series data of suspended sediment concentration with non-water body attribute status identifiers in the attribute status identifier sequence are removed to obtain qualified remote sensing inversion data of the potential collapse area.

[0022] In one embodiment, the data quality index includes a data integrity ratio; determining the suspended sediment concentration of the potential landslide area at each detection time based on the correlation between the total remote sensing observation data of the potential landslide area within the preset detection period and the qualified remote sensing inversion data includes:

[0023] The first data volume of qualified remote sensing inversion data and the second data volume of total remote sensing observation data of the potential collapse area within the preset detection period are determined.

[0024] The ratio of the first data volume to the second data volume is used as the data integrity ratio.

[0025] In one embodiment, the data quality metric includes a data validity index; after using the ratio of the first data volume to the second data volume as the data integrity ratio, the method further includes:

[0026] By fusing the data integrity ratio and the first data volume, the initial data validity index of the target suspended sediment concentration time series data of the potential collapse area at each detection time is obtained.

[0027] The data validity index is obtained by processing all the initial data validity indices.

[0028] In one embodiment, the process of processing all initial data validity indices to obtain the data validity index includes:

[0029] Extract the largest initial data validity index from all the initial data validity indices;

[0030] Based on the largest initial data validity index, all initial data validity indices are normalized to obtain the data validity index.

[0031] In one embodiment, the step of detecting landslides on the target riverbank based on the data quality indicators includes one of the following:

[0032] Determine the trend change characteristics of the data integrity ratio within the preset detection period, and if the trend change characteristics meet the preset change characteristic conditions, determine that the target riverbank has collapsed;

[0033] If the data validity index is greater than a preset data validity index threshold, it is determined that the target riverbank has collapsed.

[0034] If the trend of the ratio changes meets the preset change characteristics and the data validity index is greater than the preset data validity index threshold, it is determined that the target riverbank has collapsed.

[0035] In one embodiment, determining that the target riverbank has collapsed when the ratio trend change characteristics meet preset change characteristic conditions and the data validity index is greater than a preset data validity index threshold includes:

[0036] Obtain a first collapse assessment value derived from the ratio trend change characteristics and a second collapse assessment value derived from the data validity index;

[0037] The total collapse assessment value of the target riverbank is determined based on the first collapse assessment value, the second collapse assessment value, the first assessment weight corresponding to the first collapse assessment value, and the second assessment weight corresponding to the second collapse assessment value.

[0038] If the total collapse assessment value is greater than the preset collapse assessment threshold, it is determined that the target riverbank has collapsed.

[0039] In one embodiment, after detecting landslides on the target riverbank based on the data quality indicators, the method further includes:

[0040] Based on the data quality indicators, from the qualified remote sensing inversion data, locate the time series data segment of suspended sediment concentration that provides feedback on the detection result of the collapse of the target riverbank;

[0041] Based on the time-series data segment of suspended sediment concentration, the time period of the collapse of the target riverbank is output.

[0042] Secondly, this application also provides a riverbank collapse detection device based on remote sensing inversion time series data, comprising:

[0043] The acquisition module is used to acquire remote sensing inversion time series data of the target riverbank formed within a preset detection period, wherein the remote sensing inversion time series data includes time series data of suspended sediment concentration corresponding to each of the multiple pixel areas covered by the target riverbank.

[0044] The identification module is used to identify potential collapse areas from the multiple pixel areas based on all time-series data of suspended sediment concentration.

[0045] The filtering module is used to filter out qualified remote sensing inversion data of the potential collapse area from the time series data of the target suspended sediment concentration based on the spatiotemporal variation characteristics of the time series data of the target suspended sediment concentration in the potential collapse area.

[0046] The determination module is used to determine the data quality index of suspended sediment concentration in the potential collapse area at each detection time based on the correlation between the total remote sensing observation data of the potential collapse area during the preset detection period and the qualified remote sensing inversion data.

[0047] The detection module is used to detect the collapse of the target riverbank based on the data quality indicators.

[0048] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0049] The process involves acquiring remote sensing inversion time-series data of the target riverbank formed within a preset detection period. This data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. Based on all the suspended sediment concentration time-series data, potential landslide areas are identified from these pixel areas. According to the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data for the potential landslide areas, qualified remote sensing inversion data for the potential landslide areas is selected from the target suspended sediment concentration time-series data. Based on the correlation between the total remote sensing observation data of the potential landslide areas within the preset detection period and the qualified remote sensing inversion data, a data quality index for the suspended sediment concentration of the potential landslide areas at each detection time is determined. Finally, landslide detection is performed on the target riverbank based on the data quality index.

[0050] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0051] The process involves acquiring remote sensing inversion time-series data of the target riverbank formed within a preset detection period. This data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. Based on all the suspended sediment concentration time-series data, potential landslide areas are identified from these pixel areas. According to the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data for the potential landslide areas, qualified remote sensing inversion data for the potential landslide areas is selected from the target suspended sediment concentration time-series data. Based on the correlation between the total remote sensing observation data of the potential landslide areas within the preset detection period and the qualified remote sensing inversion data, a data quality index for the suspended sediment concentration of the potential landslide areas at each detection time is determined. Finally, landslide detection is performed on the target riverbank based on the data quality index.

[0052] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0053] The process involves acquiring remote sensing inversion time-series data of the target riverbank formed within a preset detection period. This data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. Based on all the suspended sediment concentration time-series data, potential landslide areas are identified from these pixel areas. According to the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data for the potential landslide areas, qualified remote sensing inversion data for the potential landslide areas is selected from the target suspended sediment concentration time-series data. Based on the correlation between the total remote sensing observation data of the potential landslide areas within the preset detection period and the qualified remote sensing inversion data, a data quality index for the suspended sediment concentration of the potential landslide areas at each detection time is determined. Finally, landslide detection is performed on the target riverbank based on the data quality index.

[0054] The aforementioned riverbank collapse detection method based on remote sensing inversion time-series data first acquires remote sensing inversion time-series data of the target riverbank formed within a preset detection period. This data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. In other words, remote sensing inversion technology can be used to first obtain the time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. Then, based on all the suspended sediment concentration time-series data, potential collapse areas are identified from multiple pixel areas, thus enabling the detection of riverbank collapses using suspended sediment... The purpose of using time-series concentration data is to initially conduct qualitative detection of potential riverbank collapses. Then, based on the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data in the potential collapse area, qualified remote sensing inversion data for the potential collapse area is selected from the target suspended sediment concentration time-series data. That is, by utilizing spatiotemporal variation characteristics, qualified remote sensing inversion data that can effectively invert suspended sediment concentration can be accurately selected from the target suspended sediment concentration time-series data of the potential collapse area. Furthermore, based on the total remote sensing observation data of the potential collapse area within the preset detection period and the qualified remote sensing inversion data... By establishing correlations between data points and determining the data quality index for suspended sediment concentration at each detection time in potential landslide areas, the quality of suspended sediment concentration in potential landslide areas at each detection time can be accurately reflected through data quality indicators. Ultimately, by using data quality indicators to detect target riverbanks, the aim is to achieve the goal of using data quality indicators as a reliable basis for detecting target riverbank landslides and quantitatively detecting the landslide situation. Since suspended sediment concentration can reflect the dynamic changes in the target riverbank water environment, and the data quality indicators are determined based on qualified remote sensing inversion data selected from spatiotemporal variation characteristics, they can objectively and reliably reflect the effectiveness and completeness of suspended sediment concentration in potential landslide areas at each detection time. Therefore, by using data quality indicators to detect riverbank landslides, the actual landslide situation can be reflected. This overcomes the technical deficiency that riverbank landslide detection cannot reflect the actual landslide situation due to the susceptibility of riverbank geomorphological characteristics to environmental interference and spatiotemporal changes, thus leading to false landslides. Therefore, the accuracy of riverbank landslide detection is improved. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a flowchart illustrating a riverbank collapse detection method based on remote sensing inversion time-series data in one embodiment.

[0057] Figure 2 This is a flowchart illustrating a riverbank collapse detection method based on remote sensing inversion time-series data in another embodiment;

[0058] Figure 3 This is a schematic diagram of a scenario for detecting landslides on a target riverbank in another embodiment;

[0059] Figure 4 This is a structural block diagram of a riverbank collapse detection device based on remote sensing inversion time series data in one embodiment;

[0060] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0062] First, it should be understood that riverbank collapses are usually the result of the combined effects of water flow, riverbank soil structure, and hydrological conditions. With the development of hydraulic engineering technology, the water-sediment relationship in the lower reaches of rivers has changed, exacerbating river erosion and leading to an increase in riverbank collapses. Therefore, riverbank collapse detection is crucial. Currently, riverbank collapse detection is typically carried out using the following methods: 1) Manual inspection and visual recording: Detection personnel walk or travel by boat along the riverbank, visually identifying and photographing the location and scale of collapses; 2) Image difference method: Identifying collapse areas by comparing differences in remote sensing images of the same riverbank area generated at different times; 3) Visual interpretation method: Identifying collapses by visually interpreting high-resolution remote sensing images; 4) Topographical identification method: Using elevation data to identify steep riverbanks, thereby indirectly inferring the risk of riverbank collapse; 5) Model simulation method: Simulating the riverbank collapse process by coupling hydrodynamic, sediment transport, and slope erosion models. However, each of the above detection methods has its own limitations, specifically… The first method is time-consuming, labor-intensive, and costly, and the results are affected by the subjective factors of the inspectors. The second method struggles to distinguish between permanent geomorphological changes and temporary hydrological changes, which can easily lead to false alarms of riverbank collapse. The third method relies on high-resolution data and the experience of the interpreters for accuracy, resulting in high costs and low efficiency, and it is difficult to conduct large-scale and long-term automatic monitoring. The fourth method cannot monitor the occurrence of riverbank collapse events in real time and cannot distinguish between natural slope changes and actual riverbank collapse. The fifth method requires a large amount of field data as input, and the simulation process is relatively complex, making it difficult to apply to real-time monitoring of large-scale areas. Therefore, in summary, riverbank geomorphological characteristics are easily affected by environmental interference and spatiotemporal changes. Traditional qualitative identification of riverbank geomorphological characteristics can lead to riverbank collapse detection methods that fail to reflect the actual collapse situation. Therefore, there is an urgent need for a riverbank collapse detection method based on remote sensing inversion time-series data that can improve the accuracy of riverbank collapse detection.

[0063] In one embodiment, such as Figure 1As shown, a method for detecting riverbank collapse based on remote sensing inversion time-series data is provided. This embodiment uses the method applied to a terminal as an example. The terminal includes, but is not limited to, personal computers, laptops, smartphones, and tablets. The terminal may include an acquisition module, an identification module, a filtering module, a determination module, and a detection module. The acquisition module acquires remote sensing inversion time-series data of the target riverbank formed within a preset detection period. The remote sensing inversion time-series data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. The identification module identifies potential collapse areas from multiple pixel areas based on all suspended sediment concentration time-series data. The filtering module filters out potential collapse areas from the target suspended sediment concentration time-series data based on the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data. The qualified remote sensing inversion data of the potential landslide area is used to determine the data quality index of the suspended sediment concentration in the potential landslide area at each detection time based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of the potential landslide area within a preset detection period. The detection module is used to conduct landslide detection on the target riverbank based on the data quality index. Since the suspended sediment concentration can reflect the dynamic changes of the water environment of the target riverbank, and the data quality index is determined based on the qualified remote sensing inversion data obtained by filtering spatiotemporal change characteristics, it can objectively and reliably reflect the validity and completeness of the suspended sediment concentration in the potential landslide area at each detection time. Therefore, by conducting landslide detection on the riverbank through the data quality index, it can reflect the actual landslide situation of the riverbank, thus improving the accuracy of riverbank landslide detection. It is understood that this method can also be applied to a server, and can also be applied to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps 202 to 210. Wherein:

[0064] Step 202: Obtain remote sensing inversion time series data of the target riverbank formed within a preset detection period. The remote sensing inversion time series data includes time series data of suspended sediment concentration corresponding to each of the multiple pixel areas covered by the target riverbank.

[0065] It should be noted that the target riverbank refers to the riverbank section that needs to be detected for collapse, which can be specifically defined as the spatial extraction boundary of remote sensing data; the preset detection period represents a pre-set continuous time interval for acquiring remote sensing data and conducting time-series analysis, which can be 1 year, 2 years, or 10 years, etc.; the remote sensing inversion time-series data refers to the dataset arranged in chronological order, which is obtained by inversion calculation from remote sensing images (satellite images or UAV images, etc.) within the preset detection period. Specifically, the remote sensing inversion time-series data includes the time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. A pixel area refers to the area composed of the smallest independent data unit corresponding to the geographical range of the target riverbank in the remote sensing image. The suspended sediment concentration time-series data refers to the set of suspended sediment concentration values ​​obtained by remote sensing inversion and recorded continuously in chronological order.

[0066] As an example, step 202 includes: acquiring remote sensing inversion time series data of the target riverbank formed within a preset detection period.

[0067] In one feasible approach, the riverbank requiring bank collapse detection can be identified as the target riverbank, and long-term (usually no less than one year) suspended sediment concentration data of the target riverbank can be obtained. The suspended sediment concentration data can be based on satellite imagery such as MODIS, Landsat, or Sentinel, and can be pre-generated using existing high-precision remote sensing inversion algorithms. Specifically, 40 years of suspended sediment concentration time-series data of the target riverbank can be obtained.

[0068] Step 204: Based on all time-series data of suspended sediment concentration, identify potential collapse areas from multiple pixel regions.

[0069] It should be noted that potential collapse areas refer to areas where riverbank collapses may occur. These areas can be identified through preset recognition rules. For example, in one feasible approach, a fixed suspended sediment concentration threshold can be preset. Then, the suspended sediment concentration values ​​in the time series data are compared with the preset suspended sediment concentration threshold. If a suspended sediment concentration value in a certain time series data is greater than the preset suspended sediment concentration threshold, then the pixel area corresponding to that time series data is identified as a potential collapse area.

[0070] As an example, step 204 includes: obtaining a pre-set suspended sediment concentration threshold, comparing all suspended sediment concentration values ​​in the suspended sediment concentration time series data with the suspended sediment concentration threshold one by one, and identifying the pixel areas corresponding to the suspended sediment concentration time series data where the suspended sediment concentration value is greater than the suspended sediment concentration threshold as potential collapse areas.

[0071] Step 206: Based on the spatiotemporal variation characteristics of the target suspended sediment concentration time series data of the potential collapse area, select qualified remote sensing inversion data of the potential collapse area from the target suspended sediment concentration time series data.

[0072] It should be noted that spatiotemporal variation characteristics characterize the temporal and spatial variation patterns of suspended sediment concentration time series data. Specifically, these can include temporal variation characteristics and spatial variation characteristics. Temporal variation characteristics can be the relationship between suspended sediment concentration values ​​and time, while spatial variation characteristics can be attribute status identifiers. Qualified remote sensing inversion data refers to suspended sediment concentration time series data that can truly reflect the actual situation of suspended sediment concentration in the water body of potential landslide areas. Target suspended sediment concentration time series data refers to the target suspended sediment concentration time series data of potential landslide areas.

[0073] As an example, step 206 includes: based on the temporal variation characteristics of the target suspended sediment concentration time series data of the potential collapse area, selecting suspended sediment concentration time series data with timestamps within a preset valid time series range as qualified data for remote sensing inversion.

[0074] Step 208: Based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of the potential collapse area within the preset detection period, determine the data quality index of the suspended sediment concentration in the potential collapse area at each detection time.

[0075] It should be noted that the total remote sensing observation data refers to the original data of suspended sediment concentration inversion for all pixels covering the potential landslide area within the preset detection period. It can be understood that by analyzing the correlation between the total remote sensing observation data and the qualified remote sensing inversion data, data quality indicators can be constructed. The correlation can specifically be the quantity ratio relationship and the data distribution relationship, etc. The data quality indicators characterize the data quality of suspended sediment concentration at each detection time.

[0076] As an example, step 208 includes: determining the data quality index of suspended sediment concentration in the potential collapse area at each detection time based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of the potential collapse area within a preset detection period.

[0077] Step 210: Based on the data quality indicators, conduct collapse detection on the target riverbank.

[0078] It should be noted that since the data quality index quantifies the data quality of the time-series data on the suspended sediment concentration of the target riverbank, the data quality index can reflect the collapse status of the target riverbank. For example, in one feasible approach, if the data quality index is detected to be greater than the preset data quality index threshold, it is determined that the target riverbank has collapsed; if the data quality index is detected to be less than or equal to the preset data quality index, it is determined that the target riverbank has not collapsed.

[0079] As an example, step 210 includes: determining that the target riverbank has collapsed if the detected data quality index is greater than a preset data quality index threshold, and determining that the target riverbank has not collapsed if the detected data quality index is less than or equal to the preset data quality index.

[0080] The aforementioned riverbank collapse detection method based on remote sensing inversion time-series data first acquires remote sensing inversion time-series data of the target riverbank formed within a preset detection period. This data includes time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. In other words, remote sensing inversion technology can be used to first obtain the time-series data of suspended sediment concentration corresponding to multiple pixel areas covered by the target riverbank. Then, based on all the suspended sediment concentration time-series data, potential collapse areas are identified from multiple pixel areas, thus enabling the detection of riverbank collapses using suspended sediment... The purpose of this study is to initially use time-series concentration data to qualitatively detect potential riverbank collapses. Then, based on the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data for potential collapse areas, qualified remote sensing inversion data for these areas is selected. In other words, by utilizing spatiotemporal variation characteristics, qualified remote sensing inversion data that can effectively invert suspended sediment concentrations can be accurately obtained from the target suspended sediment concentration time-series data for potential collapse areas. Finally, based on the total remote sensing observation data and the number of qualified remote sensing inversions for the potential collapse area within the preset detection period... Based on the correlation between the data and sediment concentration, the data quality index of the potential landslide area at each detection time is determined. This allows the data quality index to accurately reflect the quality of the suspended sediment concentration in the potential landslide area at each detection time. Finally, the target riverbank is monitored using the data quality index, thus achieving the goal of using the data quality index as a reliable basis for detecting riverbank landslides and quantitatively detecting the landslide situation. Since the suspended sediment concentration reflects the dynamic changes in the target riverbank's aquatic environment, and the data quality index is determined based on qualified remote sensing inversion data selected from spatiotemporal variation characteristics, it can objectively and reliably reflect the effectiveness and completeness of the suspended sediment concentration in the potential landslide area at each detection time. Therefore, by using the data quality index to detect riverbank landslides, the actual landslide situation can be reflected. This overcomes the technical defect that riverbank landslide detection cannot reflect the actual landslide situation due to the susceptibility of riverbank topography to environmental interference and spatiotemporal changes, leading to false alarms. Therefore, the accuracy of riverbank landslide detection is improved.

[0081] In one embodiment, such as Figure 2 As shown, based on all time-series data of suspended sediment concentration, potential landslide areas were identified from multiple pixel regions, including:

[0082] Step 302: Extract multiple suspended sediment concentration values ​​corresponding to each pixel region from all time-series data of suspended sediment concentration.

[0083] As an example, step 302 includes: extracting multiple suspended sediment concentration values ​​corresponding to each pixel from all time-series data of suspended sediment concentration.

[0084] Step 304: Based on all the time-series data of suspended sediment concentration, determine the suspended sediment concentration threshold corresponding to each of the multiple pixel regions.

[0085] It should be noted that since bank collapse events are accompanied by the collapse of a large amount of soil into the water, which can lead to a significant increase in local suspended sediment concentration data in a short period of time, it is possible to first determine the suspended sediment concentration threshold corresponding to each of the multiple pixel areas, and then determine whether the pixel area is a potential collapse area based on the relationship between the suspended sediment concentration value and the actual suspended sediment concentration value.

[0086] As an example, step 304 includes: determining the suspended sediment concentration threshold corresponding to each of the multiple pixel regions based on all time-series data of suspended sediment concentration.

[0087] Step 306: Based on the relationship between all suspended sediment concentration values ​​and suspended sediment concentration thresholds in each pixel area, potential collapse areas are identified from multiple pixel areas.

[0088] As an example, step 306 includes: for each pixel region, if a suspended sediment concentration value greater than a suspended sediment concentration threshold is detected among all suspended sediment concentration values ​​in that pixel region, then that pixel region is identified as a potential landslide region; if no suspended sediment concentration value greater than the suspended sediment concentration threshold is detected among all suspended sediment concentration values ​​in that pixel region, then that pixel region is not identified as a potential landslide region. In this way, by using the suspended sediment concentration threshold corresponding to each pixel region, the changes in suspended sediment in each pixel region within a preset detection period can be accurately quantified and identified, thereby reliably identifying potential landslide regions and laying the foundation for improving the accuracy of riverbank landslide detection.

[0089] In one embodiment, based on all time-series data of suspended sediment concentration, determining the suspended sediment concentration thresholds for each of multiple pixel regions includes:

[0090] The time-series data of suspended sediment concentration in each pixel region are sorted according to the magnitude of the suspended sediment concentration value to obtain the target suspended sediment concentration time-series data arranged sequentially in each pixel region; based on the effective data volume and preset percentile of the target suspended sediment concentration time-series data, the target position of the preset percentile in the target suspended sediment concentration time-series data is determined; the target suspended sediment concentration value corresponding to the target position is located in the target suspended sediment concentration time-series data, and the target suspended sediment concentration value is used as the suspended sediment concentration threshold for each pixel region.

[0091] It should be noted that a percentile positioning algorithm can be used to calculate the suspended sediment concentration threshold in real time. Then, if the suspended sediment concentration value of a certain pixel area is greater than the calculated suspended sediment concentration threshold, the pixel area is identified as a potential collapse area.

[0092] As an example, for the time-series data of suspended sediment concentration in each pixel region, all suspended sediment concentration values ​​in the time-series data are sorted in ascending order to obtain the target suspended sediment concentration time-series data for each pixel region arranged in sequence. By inputting the effective data volume and preset percentile in the target suspended sediment concentration time-series data into the preset position calculation formula, the target position of the preset percentile in the target suspended sediment concentration time-series data is calculated. The target suspended sediment concentration value corresponding to the target position is located in the target suspended sediment concentration time-series data, and the target suspended sediment concentration value is used as the suspended sediment concentration threshold for each pixel region.

[0093] In one feasible approach, time-series data of suspended sediment concentration is extracted from each pixel region for analysis. The 95th percentile of the suspended sediment concentration value in the entire suspended sediment concentration data for that pixel region is selected as the threshold. Specifically, assuming that 48 suspended sediment concentration values ​​for pixel region A (potential landslide area) in the qualified remote sensing inversion data of year a-a+1 are obtained from the overall remote sensing time-series inversion data, the 95th percentile, i.e., more than 95% of the 48 suspended sediment concentration values, are less than this value, and only 5% of the suspended sediment concentration values ​​exceed this value. That is, firstly, the 48 suspended sediment concentration values ​​are arranged in ascending order, and then the target position is obtained according to a preset position determination formula, whereby the preset position determination formula is as follows:

[0094]

[0095] Where rn refers to the target position, n is the total number of suspended sediment concentration values ​​(effective data volume), and p is the preset percentage position, which can be 95. Assuming that rn is obtained as 45.65, and assuming that the suspended sediment concentration value at the 45th position is 265 mg / L and the suspended sediment concentration value at the 46th position is 270 mg / L, the target suspended sediment concentration value at the 45.65th position can be obtained as 268 mg / L through the difference calculation method. Finally, the target suspended sediment concentration value of 268 mg / L is taken as the target suspended sediment concentration value.

[0096] In this embodiment, by locating the target position of the preset percentile in the time series data of the target suspended sediment concentration, the target suspended sediment concentration value corresponding to the target position is directly used as the suspended sediment concentration threshold for each pixel area. Since the suspended sediment concentration threshold is a personalized benchmark value calculated based on the time series data of each pixel, it can accurately match the differences in the water environment of different pixel areas, thereby providing a basis for accurately identifying potential collapse areas and laying the foundation for improving the accuracy of riverbank collapse detection.

[0097] In one embodiment, the spatiotemporal variation characteristics include attribute status identifiers; based on the spatiotemporal variation characteristics of the target suspended sediment concentration time series data of the potential landslide area, qualified remote sensing inversion data of the potential landslide area are selected from the target suspended sediment concentration time series data, including:

[0098] Based on the attribute status identifiers, an attribute status identifier sequence is generated for the target suspended sediment concentration time series data at all detection times of suspended sediment concentration. From the target suspended sediment concentration time series data, suspended sediment concentration time series data with non-water body attribute status identifiers are removed from the attribute status identifier sequence to obtain qualified remote sensing inversion data for potential landslide areas.

[0099] As an example, due to the influence of seasonal water level changes in the river channel, some pixel areas of the target riverbank may become exposed, causing some pixels of the target riverbank to change from water bodies to non-water bodies (such as sandbars) at certain times. The suspended sediment concentration data obtained at this time is obviously invalid. Therefore, by using attribute status identifiers, from the dimension of spatial change characteristics, unqualified remote sensing inversion data can be filtered out from the time series data of target suspended sediment concentration, so as to obtain qualified remote sensing inversion data of potential landslide areas.

[0100] As an example, attribute status identifiers are extracted from the target suspended sediment concentration data at each detection time. These attribute status identifiers are then arranged according to the detection time to obtain an attribute status identifier sequence. From the target suspended sediment concentration time series data, suspended sediment concentration time series data with non-water body attribute status identifiers are removed, resulting in qualified remote sensing inversion data for potential landslide areas. This allows for the removal of suspended sediment concentration time series data with non-water body attribute status identifiers, thus preserving qualified remote sensing inversion data for potential landslide areas and further laying the foundation for improving the accuracy of riverbank landslide detection.

[0101] In one embodiment, the data quality indicators include the data integrity ratio; based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of the potential landslide area within a preset detection period, the data quality indicators of the suspended sediment concentration in the potential landslide area at each detection time are determined, including:

[0102] The first data volume of qualified remote sensing inversion data and the second data volume of total remote sensing observation data of potential collapse areas within a preset detection period are determined; the ratio of the first data volume and the second data volume is used as the data integrity ratio.

[0103] It should be noted that the integrity ratio refers to the ratio within the preset detection period. [ T 0 , T n ] Within, the time t from the potential riverbank collapse detection to the final detection time point. The ratio of qualified remote sensing inversion data to total remote sensing observation data in a potential landslide area, where the integrity ratio can be expressed as: , t ∈ [ T 0 , T n ] , The calculation formula is as follows:

[0104]

[0105] in, The integrity ratio, This is the first data volume for retrieving qualified data from remote sensing inversion. This is the second largest data volume in the total remote sensing observation data.

[0106] As an example, a first data volume of qualified remote sensing inversion data and a second data volume of remote sensing observation data of potential landslide areas within a preset detection period are obtained by statistical analysis. The ratio of the first and second data volumes is used as a data quality index for the suspended sediment concentration in the potential landslide area at each detection time. In this way, landslide detection of riverbanks can be performed using the integrity ratio, reflecting the actual landslide situation and thus improving the accuracy of riverbank landslide detection.

[0107] In one feasible approach, it is assumed that... A value of 0.8 indicates that 80% of the suspended sediment concentration data is valid and 20% is invalid within the preset detection period. This allows for an accurate conclusion on the reliability of the current suspended sediment concentration time series data.

[0108] In one embodiment, the data quality metric includes a data validity index; after using the ratio of the first data volume and the second data volume as the data integrity ratio, the method further includes:

[0109] By fusing the data integrity ratio and the first data volume, the initial data validity index of the target suspended sediment concentration time series data of the potential collapse area at each detection time is obtained; all initial data validity indices are processed to obtain the data validity index.

[0110] It should be noted that after obtaining the data integrity ratio, in order to avoid the limitation that the integrity ratio alone cannot reflect the absolute number of qualified remote sensing inversions, such as the inability to further exclude the temporal anomalies in the concentration of temporary suspended sediment caused by seasonality or floods, the data validity index can be further calculated. Then, relying on the data integrity ratio and the data validity index, the collapse detection of the target riverbank can be carried out together.

[0111] As an example, the data integrity ratio and the first data volume are input into the preset validity index calculation formula to calculate the initial data validity index of the time series data of suspended sediment concentration in the potential landslide area at each detection time. The preset validity index calculation formula is as follows:

[0112]

[0113] in, This serves as the initial data validity index. The integrity ratio, The initial data volume is the first data volume for remote sensing inversion to obtain qualified data; the initial data validity index is directly used as the data validity index.

[0114] In this embodiment, by obtaining a data validity index, the shortcomings of relying solely on the integrity ratio to determine whether the amount of data is sufficient to support the analysis are not addressed, and the insufficiency of looking only at the first data volume cannot accurately measure the data quality is also compensated for, thereby further improving the accuracy of riverbank collapse detection.

[0115] In one embodiment, all initial data validity indices are processed to obtain data validity indices, including:

[0116] Extract the largest initial data validity index from all initial data validity indices; based on the largest initial data validity index, normalize all initial data validity indices to obtain the data validity index.

[0117] It should be noted that, in order to eliminate the problem of large numerical range and incomparability of the initial data validity index at different detection times due to the difference in the absolute amount of qualified data, different initial data validity indices can be uniformly mapped to the same standardized interval. This makes the threshold setting for bank collapse detection more intuitive and uniform. Furthermore, the initial validity index can be normalized to obtain a data validity index that can be compared horizontally across time series and has a clearer numerical meaning.

[0118] As an example, the largest initial data validity index is extracted from all initial data validity indices; based on the largest initial data validity index, all initial data validity indices are normalized to obtain the data validity index.

[0119] In one feasible manner, upon obtaining After that, you can The data validity index is obtained by normalization, and the specific expression for the calculation is as follows:

[0120]

[0121] in, This serves as the initial data validity index. This is a data validity index. This represents the maximum initial data validity index. Understandably, the data validity index, calculated through normalization, comprehensively considers both the integrity ratio and the initial amount of qualified data retrieved from remote sensing. This avoids the abnormal amplification of the integrity ratio at the end of the observation period for the target riverbank due to the reduction in the initial amount of total remote sensing data. This prevents the inability to accurately distinguish between real, irreversible bank collapses and temporary flooding events, ultimately leading to false alarms. This further lays the foundation for improving the accuracy of bank collapse detection for target riverbanks.

[0122] In one embodiment, the target riverbank collapse detection is performed based on data quality indicators, including one of the following:

[0123] Determine the trend of data integrity ratio within a preset detection period. If the trend of ratio change meets the preset change characteristic conditions, determine that the target riverbank has collapsed. If the data validity index is greater than the preset data validity index threshold, determine that the target riverbank has collapsed. If the trend of ratio change meets the preset change characteristic conditions and the data validity index is greater than the preset data validity index threshold, determine that the target riverbank has collapsed.

[0124] In this embodiment, during the detection of target riverbank collapse, different detection methods can be adopted to meet different scenario requirements. That is, riverbank collapse detection can be completed by relying on a single data quality indicator or by integrating multiple data quality indicators. It is understood that in the process of integrating data integrity ratio and data validity index for riverbank collapse detection, the irreversible physical process of riverbank collapse can be captured more reliably, and interference caused by insufficient data at the end of the detection can be effectively avoided, providing solid technical support for the automated monitoring of riverbank collapse.

[0125] As an example, the trend of the data integrity ratio within a preset detection period is determined. If the trend of the ratio shows a significant upward trend, the riverbank is determined to have collapsed. If the data validity index is greater than a preset data validity index threshold, the riverbank is determined to have collapsed. If the trend of the ratio shows a significant upward trend and the data validity index is greater than a preset data validity index threshold, the riverbank is determined to have collapsed.

[0126] In one embodiment, determining that a riverbank collapse exists when the ratio trend change characteristics meet preset change characteristic conditions and the data validity index is greater than a preset data validity index threshold includes:

[0127] Obtain the first collapse assessment value converted from the ratio trend change characteristics and the second collapse assessment value converted from the data validity index; determine the total collapse assessment value of the target riverbank based on the first collapse assessment value, the second collapse assessment value, the first assessment weight corresponding to the first collapse assessment value, and the second assessment weight corresponding to the second collapse assessment value; if the total collapse assessment value is greater than the preset collapse assessment threshold, it is determined that the target riverbank has collapsed.

[0128] It should be noted that, to avoid the one-sidedness of relying on a single indicator to determine whether a target riverbank has collapsed, a weighting mechanism can be set to balance the contribution of different data quality indicators to riverbank collapse detection, thereby making the results of the collapse judgment more objective. That is, corresponding weights can be set for the first collapse assessment value obtained by converting the ratio trend change characteristics and the second collapse assessment value obtained by converting the data validity index, so as to achieve a multi-dimensional comprehensive judgment. The sum of the first assessment weight and the second assessment weight is 1. For example, in one feasible method, the first assessment weight can be set to 0.6 and the second assessment weight to 0.4.

[0129] As an example, the ratio trend change characteristics are mapped to a first collapse assessment value, and the data validity index is mapped to a second collapse assessment value. By fusing the first collapse assessment value, the second collapse assessment value, the first assessment weight of the first collapse assessment value, and the second assessment weight of the second collapse assessment value, the total collapse assessment value of the target riverbank is obtained. The specific formula for fusion is as follows:

[0130]

[0131] in, This is the total collapse assessment value. This is the first collapse assessment value. This is the second collapse assessment value. As the first evaluation weight, The second assessment weight is used; if the total collapse assessment value exceeds the preset collapse assessment threshold, it is determined that the target riverbank has collapsed. In this way, by setting a weighted assessment mechanism, the core judgment role of the ratio trend is highlighted while also ensuring the basic validity of the data. This effectively reduces the occurrence of misjudgments based on a single indicator, thus improving the accuracy of collapse detection for the target riverbank.

[0132] In one embodiment, after detecting landslides on the target riverbank based on data quality indicators, the method further includes:

[0133] Based on data quality indicators, the time series data segment of suspended sediment concentration that indicates the presence of a landslide on the target riverbank is located from qualified remote sensing inversion data; based on the time series data segment of suspended sediment concentration, the time period of the landslide on the target riverbank is output.

[0134] It should be noted that after confirming that the target riverbank has collapsed, the time can be recorded to provide a basis for the subsequent work of the operators.

[0135] As an example, based on data quality indicators, the time series data segment of suspended sediment concentration that indicates the presence of a landslide on the target riverbank is located from qualified remote sensing inversion data; based on the time series data segment of suspended sediment concentration, the time period of the landslide on the target riverbank is output.

[0136] In one feasible approach, refer to Figure 3 , Figure 3This is a schematic diagram of the scenario for detecting landslides on a target riverbank. First, remote sensing inversion time-series data of the target riverbank within a preset detection period is acquired. Then, relying on all suspended sediment concentration time-series data, potential landslide areas are identified from multiple pixel regions covering the target riverbank, thus identifying potential landslide events. Next, by analyzing the spatiotemporal variation characteristics of the target suspended sediment concentration time-series data for potential landslide areas, qualified remote sensing inversion data for these areas is selected. Finally, based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data, the data quality index of the suspended sediment concentration in the potential landslide area at each detection time is determined. Finally, based on data quality indicators, the target riverbank is detected for collapse. It can be understood that the following three quantitative judgment indicators are involved in this process: 1) Suspended sediment concentration threshold: The suspended sediment concentration value exceeds its corresponding suspended sediment concentration threshold (e.g., the 95th percentile of the historical time series); 2) Data integrity ratio trend characteristics within the preset detection period: The ratio trend characteristics show a significant upward trend; 3) Data validity index: The data validity index exceeds the preset data validity index threshold (e.g., 0.9). Only when the potential collapse area meets the above three conditions simultaneously is it determined that a bank collapse event has occurred, that is, the collapse of the target riverbank is confirmed, and the time of occurrence is recorded.

[0137] Since suspended sediment concentration can reflect the dynamic changes in the target riverbank water environment, and the data quality index is determined based on qualified remote sensing inversion data obtained by screening spatiotemporal variation characteristics, it can objectively and reliably reflect the effectiveness and completeness of suspended sediment concentration in potential collapse areas at each detection time. Therefore, by using the data quality index to detect riverbank collapse, it can reflect the actual collapse situation of the riverbank. Thus, it overcomes the technical defect that riverbank collapse detection cannot reflect the actual collapse situation due to the susceptibility of riverbank geomorphological characteristics to environmental interference and spatiotemporal changes, which leads to false collapses. Therefore, it improves the accuracy of riverbank collapse detection.

[0138] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0139] Based on the same inventive concept, this application also provides a data processing device for riverbank collapse detection based on remote sensing inversion time-series data, used to implement the data processing method for riverbank collapse detection based on remote sensing inversion time-series data described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the data processing device for riverbank collapse detection based on remote sensing inversion time-series data provided below can be found in the limitations of the data processing method for riverbank collapse detection based on remote sensing inversion time-series data described above, and will not be repeated here.

[0140] In one exemplary embodiment, such as Figure 4 As shown, a data processing device for riverbank collapse detection based on remote sensing inversion time-series data is provided, comprising: an acquisition module 401, an identification module 402, a filtering module 403, a determination module 404, and a detection module 405, wherein:

[0141] The acquisition module 401 is used to acquire remote sensing inversion time series data of the target riverbank formed within a preset detection period. The remote sensing inversion time series data includes time series data of suspended sediment concentration corresponding to each of the multiple pixel areas covered by the target riverbank.

[0142] The identification module 402 is used to identify potential landslide areas from multiple pixel areas based on all time-series data of suspended sediment concentration.

[0143] The screening module 403 is used to screen qualified remote sensing inversion data of potential collapse areas from the time series data of target suspended sediment concentration based on the spatiotemporal variation characteristics of the time series data of target suspended sediment concentration in potential collapse areas.

[0144] The determination module 404 is used to determine the data quality index of suspended sediment concentration in the potential collapse area at each detection time based on the correlation between the total remote sensing observation data and the qualified remote sensing inversion data of the potential collapse area within the preset detection period.

[0145] The detection module 405 is used to detect the collapse of the target riverbank based on data quality indicators.

[0146] In one embodiment, the identification module 402 is further configured to:

[0147] Extract multiple suspended sediment concentration values ​​corresponding to each pixel region from all time-series data of suspended sediment concentration; determine the suspended sediment concentration threshold corresponding to each pixel region based on all time-series data of suspended sediment concentration; identify potential collapse areas from multiple pixel regions based on the relationship between all suspended sediment concentration values ​​and suspended sediment concentration thresholds for each pixel region.

[0148] In one embodiment, the identification module 402 is further configured to:

[0149] The time-series data of suspended sediment concentration in each pixel region are sorted according to the magnitude of the suspended sediment concentration value to obtain the target suspended sediment concentration time-series data arranged sequentially in each pixel region; based on the effective data volume and preset percentile of the target suspended sediment concentration time-series data, the target position of the preset percentile in the target suspended sediment concentration time-series data is determined; the target suspended sediment concentration value corresponding to the target position is located in the target suspended sediment concentration time-series data, and the target suspended sediment concentration value is used as the suspended sediment concentration threshold for each pixel region.

[0150] In one embodiment, the spatiotemporal change features include attribute state identifiers; the filtering module 403 is further configured to:

[0151] Based on the attribute status identifiers, an attribute status identifier sequence is generated for the target suspended sediment concentration time series data at all detection times of suspended sediment concentration. From the target suspended sediment concentration time series data, suspended sediment concentration time series data with non-water body attribute status identifiers are removed from the attribute status identifier sequence to obtain qualified remote sensing inversion data for potential landslide areas.

[0152] In one embodiment, the data quality metric includes a data integrity ratio; the determining module 404 is further configured to:

[0153] The first data volume of qualified remote sensing inversion data and the second data volume of total remote sensing observation data of potential collapse areas within a preset detection period are determined; the ratio of the first data volume and the second data volume is used as the data integrity ratio.

[0154] In one embodiment, the data quality indicators include a data validity index; the data processing device for riverbank collapse detection is also used for:

[0155] By fusing the data integrity ratio and the first data volume, the initial data validity index of the target suspended sediment concentration time series data of the potential collapse area at each detection time is obtained; all initial data validity indices are processed to obtain the data validity index.

[0156] In one embodiment, the data processing device for riverbank collapse detection is also used for:

[0157] Extract the largest initial data validity index from all initial data validity indices; based on the largest initial data validity index, normalize all initial data validity indices to obtain the data validity index.

[0158] In one embodiment, the detection module 405 is further configured to:

[0159] Determine the trend of data integrity ratio within a preset detection period. If the trend of ratio change meets the preset change characteristic conditions, determine that the target riverbank has collapsed. If the data validity index is greater than the preset data validity index threshold, determine that the target riverbank has collapsed. If the trend of ratio change meets the preset change characteristic conditions and the data validity index is greater than the preset data validity index threshold, determine that the target riverbank has collapsed.

[0160] In one embodiment, the detection module 405 is further configured to:

[0161] Obtain the first collapse assessment value converted from the ratio trend change characteristics and the second collapse assessment value converted from the data validity index; determine the total collapse assessment value of the target riverbank based on the first collapse assessment value, the second collapse assessment value, the first assessment weight corresponding to the first collapse assessment value, and the second assessment weight corresponding to the second collapse assessment value; if the total collapse assessment value is greater than the preset collapse assessment threshold, it is determined that the target riverbank has collapsed.

[0162] In one embodiment, the data processing device for riverbank collapse detection is also used for:

[0163] Based on data quality indicators, the time series data segment of suspended sediment concentration that indicates the presence of a landslide on the target riverbank is located from qualified remote sensing inversion data; based on the time series data segment of suspended sediment concentration, the time period of the landslide on the target riverbank is output.

[0164] The modules in the aforementioned data processing device for riverbank collapse detection based on remote sensing inversion time-series data can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0165] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a data processing method for riverbank collapse detection based on remote sensing inversion time-series data. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0166] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0167] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0168] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0169] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0170] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0171] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0172] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting riverbank collapse based on remote sensing inversion time-series data, characterized in that, The method includes: The remote sensing inversion time series data of the target riverbank formed within a preset detection period is obtained, wherein the remote sensing inversion time series data includes the time series data of suspended sediment concentration corresponding to each of the multiple pixel areas covered by the target riverbank. Based on all time-series data of suspended sediment concentration, potential collapse areas are identified from the multiple pixel regions. Based on the spatiotemporal variation characteristics of the target suspended sediment concentration time series data of the potential collapse area, qualified remote sensing inversion data of the potential collapse area are selected from the target suspended sediment concentration time series data. Based on the correlation between the total remote sensing observation data of the potential collapse area during the preset detection period and the qualified remote sensing inversion data, the data quality index of the suspended sediment concentration of the potential collapse area at each detection time is determined. Based on the data quality indicators, the target riverbank is subjected to collapse detection.

2. The method according to claim 1, characterized in that, The step of identifying potential landslide areas from the multiple pixel regions based on all time-series data of suspended sediment concentration includes: Extract multiple suspended sediment concentration values ​​corresponding to each pixel region from all the time-series data of suspended sediment concentration. Based on all the time-series data of suspended sediment concentration, determine the suspended sediment concentration threshold corresponding to each of the multiple pixel regions. Based on the relationship between the concentration values ​​of all suspended sediments in each pixel region and the suspended sediment concentration threshold, potential collapse areas are identified from the multiple pixel regions.

3. The method according to claim 2, characterized in that, The step of determining the suspended sediment concentration threshold corresponding to each of the multiple pixel regions based on the time-series data of all suspended sediment concentrations includes: The time-series data of suspended sediment concentration in each pixel region are sorted according to the magnitude of the suspended sediment concentration value to obtain the target suspended sediment concentration time-series data arranged sequentially in each pixel region. Based on the effective data volume and preset percentile of the time series data of the target suspended sediment concentration, determine the target position of the preset percentile in the time series data of the target suspended sediment concentration. The target suspended sediment concentration value corresponding to the target location is located in the time series data of the target suspended sediment concentration, and the target suspended sediment concentration value is used as the suspended sediment concentration threshold of each pixel region.

4. The method according to claim 1, characterized in that, The spatiotemporal variation characteristics include attribute status identifiers; the step of selecting qualified remote sensing inversion data for the potential collapse area from the time series data of the target suspended sediment concentration based on the spatiotemporal variation characteristics of the target suspended sediment concentration in the potential collapse area includes: Based on the attribute status identifiers, generate an attribute status identifier sequence for the target suspended sediment concentration time series data at all detection times of suspended sediment concentration. From the time series data of the target suspended sediment concentration, the time series data of suspended sediment concentration with non-water body attribute status identifiers in the attribute status identifier sequence are removed to obtain qualified remote sensing inversion data of the potential collapse area.

5. The method according to claim 1, characterized in that, The data quality indicators include the data integrity ratio; the determination of the suspended sediment concentration of the potential landslide area at each detection time based on the correlation between the total remote sensing observation data of the potential landslide area within the preset detection period and the qualified remote sensing inversion data includes: The first data volume of qualified remote sensing inversion data and the second data volume of total remote sensing observation data of the potential collapse area within the preset detection period are determined. The ratio of the first data volume to the second data volume is used as the data integrity ratio.

6. The method according to claim 5, characterized in that, The data quality indicators include the data validity index; After using the ratio of the first data volume to the second data volume as the data integrity ratio, the method further includes: By fusing the data integrity ratio and the first data volume, the initial data validity index of the target suspended sediment concentration time series data of the potential collapse area at each detection time is obtained. The data validity index is obtained by processing all the initial data validity indices.

7. The method according to claim 6, characterized in that, The process of processing all initial data validity indices to obtain the data validity indices includes: Extract the largest initial data validity index from all the initial data validity indices; Based on the largest initial data validity index, all initial data validity indices are normalized to obtain the data validity index.

8. The method according to claim 7, characterized in that, The step of detecting landslides on the target riverbank based on the data quality indicators includes one of the following: Determine the trend change characteristics of the data integrity ratio within the preset detection period, and if the trend change characteristics meet the preset change characteristic conditions, determine that the target riverbank has collapsed; If the data validity index is greater than a preset data validity index threshold, it is determined that the target riverbank has collapsed. If the trend of the ratio changes meets the preset change characteristics and the data validity index is greater than the preset data validity index threshold, it is determined that the target riverbank has collapsed.

9. The method according to claim 8, characterized in that, The determination that the target riverbank has collapsed when the ratio trend change characteristics meet preset change characteristic conditions and the data validity index is greater than a preset data validity index threshold includes: Obtain a first collapse assessment value derived from the ratio trend change characteristics and a second collapse assessment value derived from the data validity index; The total collapse assessment value of the target riverbank is determined based on the first collapse assessment value, the second collapse assessment value, the first assessment weight corresponding to the first collapse assessment value, and the second assessment weight corresponding to the second collapse assessment value. If the total collapse assessment value is greater than the preset collapse assessment threshold, it is determined that the target riverbank has collapsed.

10. The method according to claim 1, characterized in that, After detecting landslides on the target riverbank based on the data quality indicators, the method further includes: Based on the data quality indicators, from the qualified remote sensing inversion data, locate the time series data segment of suspended sediment concentration that provides feedback on the detection result of the collapse of the target riverbank; Based on the time-series data segment of suspended sediment concentration, the time period of the collapse of the target riverbank is output.

Images