A method for constructing morphological dynamic indices of braided channels
By constructing a braided channel morphological dynamic index based on long-term remote sensing water frequency data, the problem that existing technologies cannot quantify the dynamic evolution of braided channel planar morphology over long time scales has been solved, enabling regional-scale quantitative evaluation of braided channels and identification of high-risk river sections.
Patent Information
- Application Number
- CN202611449894.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-17
- Publication Date
- 2026-11-17
AI Technical Summary
Existing braided channel evaluation methods only focus on static spatial braiding characteristics, and cannot comprehensively quantify the degree of dynamic evolution of planar morphology over long time. They also lack a construction method for multi-dimensional index coordination and objective weighting.
Based on long-term remote sensing water frequency data, a braided channel morphological dynamic index was constructed, including objective quantitative indicators of the swing spatial range, interannual variation rate, and braiding structure complexity. The K-means clustering method was used to classify the evolution patterns.
It enables a complete depiction of the dynamic evolution characteristics of braided river morphology from both spatiotemporal dimensions, provides unified evaluation and differentiated identification at the regional scale, and supports accurate identification and engineering protection of high-risk river sections.
Smart Images

Figure CN123115840A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of river geomorphology and remote sensing geographic information technology, specifically to a method for constructing a braided channel morphological dynamic index based on long-term optical remote sensing water body frequency analysis. This method is applicable to regional-scale quantitative evaluation, evolution pattern identification, and spatial differentiation research of braided channel evolution in high-altitude, cold, and data-scarce areas. Background Technology
[0002] Braided channels are characterized by wide and shallow riverbeds, numerous sandbars, multiple distributaries, and frequent main channel shifts. Their planar morphological evolution is the result of the combined effects of hydrodynamic processes and riverbed boundary conditions. Accurately quantifying the dynamic evolution of braided channel planar morphology is crucial for river management, safety assessment of river-related projects, and regional flood risk prevention. Especially in high-altitude and cold regions, braided channels are widely distributed, accounting for approximately 81.3% of the total river length, with some sections reaching 3-5 km in width. Therefore, establishing a unified quantitative index system to measure the degree of braided channel planar morphological evolution is a prerequisite for accurate identification and engineering protection of high-risk river sections at the regional scale.
[0003] Existing methods for characterizing braided river morphology are mostly based on single-temporal remote sensing images or maps, focusing on the static description of spatial morphological complexity, such as the distributary number index, sandbar index, and distributary length index. These methods have three main shortcomings: First, they only reflect the spatial distributary pattern at a specific moment, failing to depict the dynamic migration of the river's planar position over long time scales, and failing to uniformly quantify the two evolutionary characteristics of "spatial complexity" and "temporal variability." Second, they do not incorporate the spatial range of oscillation, interannual variation frequency, and distributary structural complexity into the same characterization system, making it difficult to comprehensively reflect the multidimensional evolutionary attributes of braided river planar morphology. Third, the weighting of indicators largely relies on empirical assignment, lacking a data-driven objective weighting mechanism, which limits the comparability between multiple river segments at the regional scale.
[0004] In recent years, the accumulation of long-term optical remote sensing data (such as Landsat series archives of over 40 years) has made it possible to quantify braided channel evolution from both spatiotemporal dimensions. Water frequency maps can transform temporal morphological change information into a stable spatial probability distribution, providing a unified data foundation for simultaneously extracting the spatial extent of undulation, interannual disturbance frequency, and braiding structural complexity. However, current technologies lack a method for constructing a braided channel morphological dynamics index based on water frequency maps, multi-dimensional index coordination, and objective weighting. Therefore, it is urgent to overcome these technical bottlenecks and establish a braided channel evolution index suitable for quantitative evaluation at the regional scale. Summary of the Invention
[0005] The technical problem to be solved by this invention is to overcome the shortcomings of existing braided channel evaluation methods, which only focus on static spatial bifurcation characteristics and cannot comprehensively quantify the degree of dynamic evolution of planar morphology over long time scales. This invention provides a method for objectively constructing a braided channel morphological dynamic index based on long-term remote sensing water frequency data, from three dimensions: swing spatial range, interannual variation rate, and bifurcation structural complexity.
[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:
[0007] A method for constructing morphological dynamic indices of braided channels includes the following steps:
[0008] S1. Data preparation and preprocessing: Obtain annual remote sensing images of the target braided river segment over consecutive years T, where T ≥ 20 years; extract water information from the remote sensing images for each year, and generate an annual water binary map using a normalized water index combined with threshold segmentation or classification algorithms, assigning a value of 1 to water pixels and 0 to non-water pixels; unify the geographic coordinate system and spatial resolution of all images.
[0009] Furthermore, the annual remote sensing images are selected from Landsat series or Sentinel-2 satellite images during the annual high-water season to ensure the consistency and comparability of water body extraction; the water body index adopts the improved Normalized Difference Water Index (MNDWI), and the threshold is determined by the OTSU adaptive method or a unified threshold.
[0010] S2. Construct a water body frequency map. Based on the T-year water body binary map obtained in step S1, count the number of times each pixel in the study area is identified as a water body in the T-year period, and divide by T to obtain the water body frequency map F. The value of F ranges from 0 to 1. Wherein, F=0 indicates that the pixel has never been occupied by a water body during the observation period, and F>0 indicates that the pixel has a water body in at least one year.
[0011] The water body frequency map compresses long-term water body distribution information into a single spatial probability distribution map: a higher F value indicates that the location is stably occupied by water for a long period of time, corresponding to the main channel or stable distributary; a medium F value indicates that the location is in an active zone where the channel frequently shifts; an F value approaching 0 indicates that the location is on stable land at or outside the edge of the channel's active range. All subsequent sub-indices are derived from this same water body frequency map, ensuring the inherent consistency between indicators in terms of spatial range and temporal scale.
[0012] S3. Calculate the swing amplitude index R swing Based on the water frequency map F obtained in step S2, all pixels with water frequency values satisfying F>0 are defined as historical activity areas, representing the maximum activity range of the braided channel during the observation period. A first frequency threshold F is set. low =0.25 and the second frequency threshold Fhigh =0.75. The ratio of the total area of pixels with water frequency values between 0.25 and 0.75 to the total area of pixels in the historical activity zone is calculated; this ratio is the swing amplitude index R of that river segment. swing R swing The larger the value, the wider the spatial range involved in the river channel meandering and the more intense the meandering. R swing It is an area ratio index at the overall scale of the river section, which is calculated directly using the river section study area as the statistical unit, without the need for a spatial averaging step.
[0013] The selection of 0.25 and 0.75 is based on the fact that these two thresholds correspond to the quarter and third quartiles of the water body frequency, respectively. Statistically, they can stably characterize the reasonable range of the active swing zone of the river channel, avoid subjective threshold selection bias, and have good universality.
[0014] S4. Calculate the interannual rate of change index, Rate. change Based on the annual water body binary map time series obtained in step S1, for the pixels in the water body frequency map of this river section that satisfy F>0, their frequency is statistically analyzed. For the state transition between adjacent years, the probability P(0|1) of the transition from water body to non-water body and the probability P(1|0) of the transition from non-water body to water body are calculated for each pixel, and the sum of the two is taken as the interannual variation rate of that pixel. The interannual variation rate ranges from 0 to 2, and the larger the value, the more frequent the interannual reversal of the riverbed state at that location.
[0015] The interannual rate of change index Rate is obtained by spatially averaging the interannual change rates of all pixels within the river segment that satisfy F>0. change Rate is used to reflect the overall activity level of interannual adjustment of river boundaries at the river section scale. change The larger the value, the more frequent and active the interannual changes in the river channel boundary of that section. change The calculations are performed only on pixels within the historical activity zone to exclude the dilution effect of stable land outside the activity zone on the index.
[0016] S5. Calculate the bifurcation structure index (BFI). In the water frequency map F of the target river segment, select pixels with F>0 to construct potential river activity zones. Arrange several cross-sections at equal intervals along the longitudinal direction of the river channel. The spacing between the cross-sections is determined based on the length and width of the river segment, with a range of 50m to 100m.
[0017] At each cross-section, a water frequency map F is extracted using a fixed frequency threshold Fa=0.5. Pixels with F≥0.5 are assigned a value of 1 (considered as "active distributary water"), and pixels with F<0.5 are assigned a value of 0, resulting in a binarized water mask for that cross-section. Continuous water pixel segments on this mask are identified, representing each active distributary, and the width of each distributary (the lateral span of a continuous water pixel segment) is calculated. The proportion w of the main distributary width at that cross-section is then calculated. main = (Maximum channel width) / (Cross-sectional active zone width), where the cross-sectional active zone width is the total width of all continuous pixel segments with F>0 on the cross-section. Count the number of channels N (the number of continuous water cell segments) on the cross-section. Then, the branching structure value of the cross-section is:
[0018]
[0019] Where log represents the natural logarithm or the common logarithm. BFI i The value range is [0, +∞). The larger the value, the more complex the cross-sectional branching structure is (the smaller the proportion of the main branch and the more branches).
[0020] The physical meaning of the fixed threshold Fa=0.5 is that pixels with a water frequency greater than or equal to 0.5 are covered by water in most years, representing stable channels in a time-averaged sense. This threshold has a clear physical meaning and does not need to be dynamically adjusted.
[0021] BFI of all sections within this river section i The arithmetic mean is taken to obtain the bifurcation structure index (BFI) for this river segment. A higher BFI value indicates that the river segment has more dispersed flow, denser sandbar divisions, and a more complex bifurcation structure in a time-averaged sense. BFI is related to R... swing Rate change The directions are consistent, and the larger the value, the more active the evolution.
[0022] S6. Construct the braided channel morphological dynamics index BRMI, based on the swing amplitude index R obtained in step S3. swing The interannual rate of change index Rate obtained in step S4 change The braided channel morphodynamic index BRMI, obtained in step S5 and the number of observation years T, is calculated using the following formula:
[0023]
[0024] In the formula, a larger BRMI value indicates a more intense dynamic evolution of the braided channel's planar morphology in that river segment. The physical meaning of this formula is: [The formula is expressed as a percentage of the interannual variation rate, Rate]. change As a baseline activity level, it is amplified by two enhancement factors (by geometric mean) of oscillation amplitude and bifurcation structure, and then divided by the number of observation years to obtain the average activity level per unit time.
[0025] S7. Evolutionary pattern classification, based on R of each target river segment swing Rate change The three sub-indices, BFI, and K-means clustering were used to classify evolution patterns. The optimal number of clusters was determined jointly using the elbow rule and silhouette coefficient. The elbow rule uses the inflection point of the sum of squares within each cluster as a function of the number of clusters to determine the optimal range of cluster numbers; the silhouette coefficient uses the maximum average silhouette width for each cluster number to determine the optimal number of clusters. When the results of the two methods are inconsistent, the clustering scheme that balances the sample size of each cluster is preferred.
[0026] Based on the mean characteristics of each category across the three sub-indicators, feature descriptions are provided for each evolutionary pattern category. For example, a category with a significantly high value for a certain sub-indicator corresponds to an evolutionary pattern dominated by that feature, while categories with all three sub-indicators being high or low correspond to either a comprehensively active or relatively stable pattern, respectively. The specific number of clusters and category names are determined based on the data characteristics of the actual river section samples; this invention does not impose specific limitations on the number of clusters or category names.
[0027] S8. Evolutionary grading and spatial map generation: The natural discontinuity method is used to cluster the BRMI values of multiple target river segments, automatically determining the optimal classification breakpoints and classifying the dynamic evolution of the braided channel planar morphology into a preset number of grades. Preferably, there are 5 grades, corresponding to: relatively stable, slight evolution, moderate evolution, relatively strong evolution, and strong evolution. This evolutionary grading answers the question of "how strong is the evolution," providing an intensity scale for unified lateral comparison between different river segments.
[0028] Based on the evolution pattern category and evolution level of each river segment, the two are spatially overlaid and plotted to generate a regional-scale spatial map of the dynamic evolution of braided river planar morphology, achieving a two-dimensional collaborative diagnosis of "intensity + type". On the same map, it is possible to see both "which river segments are undergoing strong evolution" (level) and "which evolution type the river segment belongs to" (pattern), providing a direct basis for differentiated identification of high-risk river segments and engineering protection decisions.
[0029] The beneficial effects of the present invention, compared with the prior art, include the following significant advantages:
[0030] ① This method breaks through the limitations of traditional approaches that only focus on the spatial braiding morphology of a single temporal phase. For the first time, it incorporates the oscillation spatial range, interannual variation rate, and braiding structural complexity into a unified quantitative framework. All three indicators are derived from the frequency diagram of the same water body, ensuring the inherent consistency between the indicators in terms of spatial range and temporal scale. It comprehensively depicts the dynamic evolution characteristics of braided channel planar morphology from both spatiotemporal dimensions, achieving a comprehensive measurement of "spatial complexity" and "temporal variability."
[0031] ② Use a fixed threshold (R) swing F low =0.25, F high =0.75; BFI's Fa=0.5), avoiding the subjectivity and computational burden of threshold traversal and sensitivity analysis, making the method simpler and more efficient. At the same time, the threshold selection has clear statistical significance (quartiles) and physical significance (time-averaged stable channel), ensuring the stability and repeatability of the index.
[0032] ③ The branch structure index BFI adopts the product form of cross-sectional statistics (1-w) main The expression )×log(N+1) considers both the proportion of main channels and the number of channels, has a clear physical meaning, and is consistent with R. swing Rate change They are in the same direction.
[0033] ④ All three indicators use river sections as the final analysis unit, ensuring consistency in the spatial scale of the indicators. swing Rate represents the overall area ratio of this river section. change The spatial average of pixels within the river segment is given by , and the cross-sectional average of the bifurcation structure values of each section within the river segment is given by . All three reflect the overall characteristics of the river segment scale and are aligned in the direction of "the larger the value, the more active the evolution". This provides a consistent sample scale benchmark for subsequent cluster analysis and hierarchical classification.
[0034] ⑤ This invention directly constructs the BRMI using a product-square root combination formula, eliminating the need for standardization and principal component analysis weighting, thus avoiding the uncertainty of subjective weighting while maintaining the physical transparency of the comprehensive index. The BRMI formula has a simple structure and completely fixed parameters, making it easy to achieve automated batch calculations over a large area and multiple river segments.
[0035] ⑥ Based on the generation of a single comprehensive index BRMI, K-means clustering based on three sub-indices is used to classify evolutionary pattern types (highly oscillating and active type, highly structurally complex type, comprehensively active type, and relatively stable type), achieving differentiated identification of evolutionary characteristics of different types and overcoming the technical deficiency that a single comprehensive index cannot distinguish evolutionary types. Evolutionary patterns and evolutionary levels work together to fully support the dynamic evolution analysis of planar morphology from both the dimensions of type and intensity.
[0036] ⑦ This method can be automated and batch-processed based on remote sensing cloud platforms or programming. It is suitable for regional-scale evaluation of braided channel evolution over large areas and multiple river sections. It can generate spatial maps of the dynamic evolution of braided channel planar morphology in plateau rivers, directly serving practical applications such as identification of high-risk river sections in major linear engineering projects, bridge site selection, and protection design. Attached Figure Description
[0037] Figure 1The overall flowchart of the braided channel morphological dynamic index construction method provided in the embodiments of the present invention is shown.
[0038] Figure 2 The water frequency diagram and oscillation amplitude index R in the embodiment are shown. swing interannual rate of change change Calculation diagram.
[0039] Figure 3 This is a schematic diagram illustrating the river channel cross-section segmentation and the calculation of the branching structure index (BFI) in the embodiment.
[0040] Figure 4 This table shows the evolutionary hierarchy and typical landform features. Detailed Implementation
[0041] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.
[0042] like Figures 1 to 4 As shown in the figure, this embodiment takes the regional-scale morphological evolution evaluation of a typical braided river section as an example to illustrate the implementation process and application effect of the present invention.
[0043] S1. Data Preparation and Preprocessing: Taking braided channels in typical river sections as the research object. First, based on the global alluvial river model dataset and visual interpretation of multiple remote sensing images, a unified coding sample library of braided river sections was established, which was divided into several independent braided river section research units of appropriate length.
[0044] Landsat series (Landsat5 TM, Landsat7 ETM+, Landsat8 / 9 OLI) and Sentinel-2 satellite imagery for typical river sections from 1986 to 2025 were acquired from the Google Earth Engine platform. Each year, the image with the lowest cloud cover during the high-water season (July to September) was selected. After preprocessing all images including radiometric calibration and atmospheric correction, a modified Normalized Difference Water Index (MNDWI) was used with a uniform threshold to extract the maximum water body extent during the high-water season each year, generating 40 annual binary water body maps with a spatial resolution of 30 meters. Water body pixels were assigned a value of 1, and non-water body pixels were assigned a value of 0. Effective observation counts were screened for all river sections, and sections with less than 70% of effective observations were removed to ensure the reliability of the water body frequency statistics.
[0045] S2. Construct a water frequency map by overlaying 40 binary images of typical river segments. Count the number of years each pixel was occupied by water over 40 years, divide by 40 to obtain the water frequency value F, and generate the water frequency map for each river segment. A higher F value indicates that the location is stably occupied by water for a long period; a medium F value indicates that the location is in an active zone with frequent river channel shifts; and an F value approaching 0 indicates that the location is on stable land at or outside the river channel's active range. The water frequency map compresses 40 years of temporal variation information into a single spatial probability distribution map, providing a unified and comparable data foundation for the calculation of all subsequent sub-indicators.
[0046] S3. Calculate the swing amplitude index R swing On the water frequency map of each river section, all pixels with F>0 are designated as historical activity areas. F is set... low =0.25, F high =0.75. The ratio of the total area of pixels with water frequency values between 0.25 and 0.75 to the total area of historical activity zones is calculated to obtain R. swing In this embodiment, a typical river section R swing =0.54, indicating that 54% of the historical activity area in this river section belongs to the active swing zone, with a wide swing range.
[0047] S4. Calculate the interannual rate of change index, Rate. change For pixels in the water frequency maps of each river segment that satisfy F>0, based on 40 annual binary water maps, the state transitions between 39 pairs of adjacent years are analyzed. For each pixel, the number of times it transitions from water to land (1→0) and the number of times it transitions from land to water (0→1) are counted. These are divided by 39 to obtain the conditional probabilities P(0|1) and P(1|0), respectively. The sum of these two probabilities is the interannual variability rate of that pixel. The spatial average of the interannual variability rates of all pixels satisfying F>0 is taken to obtain the Rate. change In this embodiment, the river section's Rate change =0.52, indicating that there is an average of 52% probability of a land-water state reversal per pixel per year.
[0048] S5. Calculate the bifurcation structure index (BFI) and select pixels with F>0 on the water frequency map of each river segment to construct potential channel activity areas. Establish 30 cross-sections at equal intervals of 150m along the longitudinal direction of the river. On each cross-section, perform binarization with a fixed threshold Fa=0.5 to identify continuous water cell segments (bifurcations). Calculate the width of each bifurcation, identify the widest bifurcation, and calculate the proportion w of the main bifurcation width. main =Maximum channel width / Cross-sectional active zone width. Count the number of channels N. Calculate the BFI of this cross-section. i =(1-w main )×ln(N+1) (This embodiment uses the natural logarithm). BFI for all cross-sectionsi The arithmetic mean is taken to obtain the BFI for this river segment. In this embodiment, the average w main =0.45, average N=4.2, then BFI=(1-0.45)×ln(5.2)=0.55×1.649=0.91, indicating that the river section has a complex branching structure.
[0049] S6. Construct the Braided Channel Morphology Dynamics Index (BRMI) and apply the obtained R... swing =0.54, Rate change Substituting 0.52, BFI=0.91, and T=40 into the formula:
[0050]
[0051] The BRMI value of this river section is 0.0223, indicating that its dynamic evolution of planar morphology is at a moderate to high level.
[0052] S7. Evolutionary pattern classification, based on R in each river segment swing Rate change The evolution patterns of the three sub-indices (R, BFI, and ROI) were classified using K-means clustering. The elbow rule was used to analyze the trend of the sum of squares within clusters with the number of clusters k (1-8), revealing a significant inflection point at k=4. Simultaneously, the silhouette coefficient was calculated for each cluster number, with the highest silhouette coefficient (0.62) at k=4. The optimal number of clusters was determined by combining the two methods. Based on the mean characteristics of each category across the three sub-indices, typical evolution patterns were named: High Swinging Active Pattern (R...). swing Types include: (significantly higher than other types), (highly complex structure type, BFI mean significantly higher than other types), (comprehensively active type, all three indicators are higher than the overall mean), and (relatively stable type, all three indicators are lower than the overall mean).
[0053] S8. Evolution Level Classification and Spatial Map Generation: The BRMI values of each river segment are clustered using the natural discontinuity method to automatically identify natural discontinuities in the data distribution, classifying the evolution degree into five levels: relatively stable, slightly evolved, moderately evolved, relatively strong evolved, and strongly evolved. The evolution pattern type and evolution level of each river segment are spatially overlaid and plotted to generate a spatial map of the morphological dynamics evolution of braided channels. This clearly displays the two-dimensional characteristics of "intensity + type" and their spatial differentiation patterns of different braided river segments of main streams and tributaries.
[0054] The above embodiments are merely illustrative of the technical solutions of the present invention and are not intended to limit them. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for constructing morphological dynamic indices of braided channels, characterized in that, Includes the following steps: S1. Data preparation and preprocessing: Obtain annual remote sensing images of the target braided river section over a continuous period of T years, extract water information from the remote sensing images of each year, generate an annual water body binary map, and unify the geographic coordinate system and spatial resolution of all images. S2. Constructing a water body frequency map: Based on the T-year water body binary map obtained in step S1, count the number of times each pixel in the study area is identified as a water body in the T-year period, and divide by T to obtain the water body frequency map F. S3, calculate the swing amplitude index R swing : based on the water body frequency map F obtained in step S2, define all pixels with water body frequency value satisfying F>0 as historical active area; set a first frequency threshold F low and a second frequency threshold F high ; calculate the ratio of the total area of pixels with water body frequency value between F low and F high to the total area of pixels in the historical active area, and the ratio is the swing amplitude index R swing of the river section S4. Calculate the interannual rate of change index, Rate. change Based on the annual water body binary map time series obtained in step S1, for the pixels in the water body frequency map of this river segment that satisfy F>0, their frequency distribution in the frequency map is statistically analyzed. For the state transition between adjacent years, the probability P(0|1) of water body transitioning to non-water body and the probability P(1|0) of non-water body transitioning to water body are calculated for each pixel, and the sum of the two is taken as the interannual rate of change for that pixel. The interannual rate of change of all pixels in the river segment that satisfy F>0 is spatially averaged to obtain the interannual rate of change index Rate for that river segment. change ; S5. Calculate the Bifurcation Structure Index (BFI): In the target river segment's water frequency map F, select pixels with F>0 to construct potential active river channels; arrange several cross-sections at equal intervals along the longitudinal direction of the river channel; on each cross-section, extract the water frequency map F with a fixed frequency threshold Fa to obtain a binarized water mask for that cross-section; identify continuous water pixel segments on this mask as active channels, and count the width of each channel; calculate the proportion w of the main channel width in that cross-section. main = (Maximum branch width) / (Cross-section active area width), where the cross-section active area width is the total width of all continuous pixel segments with F>0 on the cross-section; count the number of branches N on the cross-section; then the branching structure value of the cross-section is BFI. i =(1-w main )×log(N+1), where log is the natural logarithm or the common logarithm; BFI for all cross-sections within this river segment. i The arithmetic mean is taken to obtain the bifurcation structure index (BFI) for this river section; S6. Construct the Braided Channel Morphological Dynamics Index (BRMI): Based on the swing amplitude index R obtained in step S3. swing The interannual rate of change index Rate obtained in step S4 change The braided channel morphodynamic index BRMI, obtained in step S5 and the number of observation years T, is calculated using the following formula: , S7. Evolutionary Pattern Classification: Based on R of each target river segment swing Rate change The three sub-indices, BFI, and K-means clustering method were used to classify the evolution patterns. S8. Evolutionary Classification and Spatial Map Generation: The natural discontinuity method is used to perform cluster analysis on the BRMI values of multiple target river segments, automatically determine the optimal classification discontinuity, and classify the dynamic evolution degree of braided river planar morphology into a preset number of levels; based on the evolutionary pattern category and evolutionary level of each river segment, a regional-scale spatial map of the dynamic evolution of braided river planar morphology is generated.
2. The method according to claim 1, characterized in that, In step S3, the first frequency threshold F low =0.25 and the second frequency threshold F high The selection of 0.75 is based on the fact that 0.25 and 0.75 correspond to the quarter and third quartiles of the water body frequency, respectively. Statistically, these values can stably characterize the reasonable range of the active swing zone of the river channel, avoiding subjective threshold selection bias.
3. The method according to claim 1, characterized in that, In step S4, the calculation of the interannual variation rate is performed only for pixels in the water frequency map that satisfy F>0, so as to exclude the dilution effect of stable land outside the historical activity area on the index.
4. The method according to claim 1, characterized in that, In step S5, the cross-sections are arranged at equal intervals, the interval being determined based on the river length and width, ranging from 50m to 100m; the width of the active zone of the cross-section is the total width of all continuous pixel segments with F>0 on that cross-section; the width of the main channel accounts for w. main = Maximum channel width / Cross-sectional active area width; The number of channels N is the number of continuous water cell segments on the cross-section; The physical meaning of the fixed frequency threshold Fa=0.5 is: cells with a water frequency greater than or equal to 0.5 are covered by water in most years, representing stable channels in the sense of time average.
5. The method according to claim 1, characterized in that, In step S6, the number of observation years T is consistent with the number of years of continuous remote sensing images in step S1, and the value range is T≥20 years.
6. The method according to claim 1, characterized in that, The method further includes: after step S6, normalizing the BRMI values of each river segment to make them range from 0 to 1, so as to facilitate horizontal comparison between different regions.
7. The method according to claim 1, characterized in that, In step S7, the optimal number of clusters is determined by combining the elbow rule and the silhouette coefficient. The elbow rule determines the range of the optimal number of clusters by the inflection point of the sum of squares within the cluster as a function of the number of clusters, and the silhouette coefficient determines the optimal number of clusters by the maximum value of the average silhouette width under each number of clusters. When the results of the two methods are inconsistent, the clustering scheme that balances the sample size of each cluster is selected.
8. The method according to claim 1, characterized in that, In step S8, the preset number of evolution levels is 5, corresponding to: relatively stable, mild evolution, moderate evolution, relatively strong evolution, and strong evolution.
