High mountain accumulated snow thickness and reserve estimation method based on remote sensing and physical constraint
By combining remote sensing and physical constraints with remote sensing indices and support vector machine classification, the problems of unstable snow cover identification and inaccurate thickness estimation in high-altitude areas were solved, and stable estimation and reliable extrapolation of high-altitude snow cover information were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHENGDU UNIVERSITY OF TECHNOLOGY
- Filing Date
- 2026-01-05
- Publication Date
- 2026-05-01
AI Technical Summary
Existing technologies lack stability in identifying snow cover in high-altitude areas. The estimation of snow thickness and reserves relies on a large amount of measured or meteorological data and lacks a unified physical constraint applicable to complex terrain, resulting in significant misjudgments and uncertainties.
Using a method based on remote sensing and physical constraints, the snow thickness and storage were calculated by initial screening with remote sensing indices, classification with support vector machines, and terrain correction, combined with a base shear stress-slope-density model. Statistical relationships between snow area and thickness, and between thickness and storage were then constructed.
It improves the stability of snow cover classification and the reliability of thickness and reserve estimation, and is applicable to high-altitude areas with limited samples and scarce observations. It reduces the false positive rate and improves the accuracy of estimation.
Smart Images
Figure CN121962234A_ABST
Abstract
Description
A method for estimating snow thickness and reserves in high mountains based on remote sensing and physical constraints Technical Field
[0001] This invention relates to a method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints, belonging to the field of snow cover identification technology. Background Technology
[0002] High-altitude snow cover is a crucial source of water resources in mountainous areas, and its area, thickness, and volume directly determine the amount of snowmelt replenishment and the availability of regional water resources. Accurately obtaining the spatial distribution, thickness, and volume of snow cover is fundamental for conducting high-altitude hydrological process analysis and water resource assessment. However, in high-altitude regions with dramatic topographic relief and complex surface types, snow cover is characterized by its small area, fragmented distribution, and rapid spatial and temporal changes, making traditional observation and inversion methods difficult to apply directly.
[0003] Existing snow cover identification technologies mainly rely on optical remote sensing indices or supervised classification methods. While these methods can provide snow cover information over a wide area, they are prone to misclassification in mountain shadows, areas with a mixture of bare rock and thin snow, leading to unstable spatial distribution identification of snow cover. Common methods for estimating snow thickness and volume include manual interpolation, empirical models, machine learning inversion, and meteorological driven models. However, most of these methods rely on dense samples or complete meteorological parameters. In high-altitude areas, the scarcity of observation stations and the poor continuity of meteorological data make these methods prone to significant spatial uncertainty in practical applications.
[0004] Furthermore, existing methods for snow thickness or reserves inversion often lack a unified constraint framework applicable to complex mountainous terrain. Snow thickness varies regularly with altitude, slope aspect, wind field, and micro-topographical conditions, but most inversion processes do not systematically incorporate these physical constraints, resulting in significant deviations in snow thickness estimations in fractured and thin-layered snow areas, which in turn affects the reliability of regional-scale snow reserves estimations.
[0005] Therefore, existing technologies lack stability in snow cover identification in high-altitude areas, and their thickness and reserve estimation rely on large amounts of data and lack physical constraint mechanisms, making it difficult to meet the needs of effectively inverting snow cover thickness and reserve in areas with weak observation. Summary of the Invention
[0006] The purpose of this invention is to address the problems in existing technologies, such as insufficient accuracy in identifying snow cover in high-altitude areas, reliance on large amounts of measured or meteorological data for estimating snow thickness and reserves, and lack of unified physical constraints applicable to complex terrain. This invention provides a method for estimating snow thickness and reserves in high-altitude areas based on remote sensing and physical constraints. It aims to achieve a relatively stable and scalable estimation process for the spatial distribution, thickness, and snowmelt reserves of high-altitude snow cover under conditions of limited samples and scarce observations. This will improve the applicability and reliability of high-altitude snow cover information inversion and provide an executable methodological basis for assessing snowmelt resources in complex mountainous areas.
[0007] The technical solution provided by this invention to solve the above-mentioned technical problems is: a method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints, comprising the following steps: Step S10: Acquire remote sensing image data of the target area at different time phases over several consecutive years, and acquire corresponding digital elevation model data; identify the snow cover area of each year's images based on remote sensing indices to obtain multi-year snow cover spatial distribution results and its patch vector data; Step S20: Based on the multi-year snow cover identification results, statistically analyze the area, spatial distribution range, and altitude range of snow cover patches for each year, extract the upper limit height, lower limit height, and interannual variation characteristics of snow cover distribution, which are used to characterize the stability of multi-year snow cover spatial distribution. And fluctuation characteristics; Step S30: Based on the obtained multi-year snow patches, classify the snow patches according to the patch area level, analyze the proportion and structural changes of snow patches of different area levels in each year, so as to depict the spatial pattern characteristics of snow patches evolving from dispersion to concentration or vice versa; Step S40: For the snow patches identified in each year, calculate the corresponding spatial distribution of snow thickness, and further combine the density constraint model to calculate the water storage of snow at the patch and regional scale; Step S50: Based on the area, thickness and storage data of multi-year snow patches, construct the statistical relationship between snow area and thickness, and thickness and storage, for the estimation of snow thickness and storage in high mountains.
[0008] A further technical solution is that the specific process of identifying snow cover area based on remote sensing index for images of different years in step S10 is as follows: Step S11: Perform radiometric calibration, atmospheric correction, cropping, and topographic correction on multi-temporal remote sensing images of the target area; Step S12: Construct a normalized snow cover index based on green light and shortwave infrared bands; Step S13: Use the normalized snow cover index threshold as an initial screening condition and mark pixels that meet the threshold as suspected snow cover areas; Step S14: Perform supervised classification on suspected snow cover areas; Step S15: Convert the classification results into vector boundaries, and manually verify the misjudged areas by combining image texture, tone, and slope aspect features, thereby obtaining the snow cover area distribution results.
[0009] A further technical solution is that the calculation formula in step S12 is:
[0010] In the formula: and These are the green light and shortwave infrared reflectance of the remote sensing image, respectively. This is the normalized snow cover index.
[0011] A further technical solution is that, in step S14, a support vector machine classification method is used for supervised classification.
[0012] A further technical solution is that the discriminant function of the support vector machine classification method is:
[0013] In the formula: x i For training samples, y i For category labels, K( is the weighting coefficient) ) is the kernel function.
[0014] A further technical solution is that the formula for calculating the spatial distribution of snow thickness in step S40 is:
[0015] In the formula: H is the snow thickness, The average base shear stress, and Here, θ represents the density of the base ice and the density of the snow, respectively, and θ is the slope. Let be the acceleration due to gravity, and a and b be coefficients.
[0016] A further technical solution is that the density constraint model in step S40 is:
[0017] In the formula: Z is the altitude, T is the near-surface temperature, and τ is the snow cover age.
[0018] A further technical solution is that the formula for calculating the reserves in step S40 is:
[0019] Where: h i For the thickness of the i-th layer of snow, and These are the densities of snow and water, respectively.
[0020] The present invention has the following beneficial effects: 1. The joint identification method of "NDSI index initial screening + SVM supervised classification + terrain feature correction" proposed in this invention can significantly reduce snow cover classification misjudgment compared with the single index method. This effect comes from the fact that the present invention incorporates spectral features and multi-source information such as morphology, slope aspect, and texture into the classification decision, making snow cover boundary identification more stable.
[0021] 2. The present invention infers thickness based on the physical constraints of "base shear stress-slope-density", which can be run with very few sampling points, indicating that the physical constraint model of the present invention can effectively avoid the extrapolation instability problem of traditional interpolation methods in the case of "patchy snow".
[0022] 3. By integrating snow cover area identification, thickness inference, and snow water equivalent extension model, a complete regional snow cover storage distribution can be generated. The method of this invention shows stable performance in snow water storage calculations across different years. This indicates that the method is applicable to typical "small-scale, highly heterogeneous, and patchy" snow cover structures, representing a significant extension of existing snow cover inversion techniques. Attached Figure Description
[0023] Figure 1 shows the cumulative and changing snow cover area over the years; Figure 2 shows the annual snow thickness and reserves; Figure 3 shows the snow cover thickness and reserves at different areas (based on 5000 m²). 2 Comparison of the relationship between thickness and reserves under snow patch conditions (with the boundary as the boundary). Detailed Implementation
[0024] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0025] This invention provides a method for estimating the thickness and storage of high-altitude snow cover based on remote sensing and physical constraints. Through snow cover area identification, thickness inference, snow water equivalent calculation, and extended estimation, it achieves a systematic quantification of high-altitude snow water storage. Specifically, it includes the following steps: Step S10: Acquire remote sensing image data of the target area over several consecutive years at different time phases, and acquire corresponding digital elevation model data. Based on the remote sensing index, identify the snow cover area of each year's image to obtain multi-year snow spatial distribution results and its patch vector data; Step S11: Perform radiometric calibration, atmospheric correction, cropping, and topographic correction processing on the multi-temporal remote sensing images of the target area; Step S12: Construct a normalized snow cover index based on green light and shortwave infrared bands.
[0026] In the formula: and These are the green light and shortwave infrared reflectance of the remote sensing image, respectively. The normalized snow cover index is used as the initial screening condition; step S13: using the normalized snow cover index threshold as the initial screening condition, pixels that meet the threshold are marked as suspected snow cover areas; step S14: supervised classification is performed on the suspected snow cover areas; specifically, a support vector machine classification method is used for supervised classification, that is, a training set is constructed using typical snow cover and non-snow cover samples, and a discriminant model is constructed using a radial basis kernel function, the discriminant function of which is:
[0027] In the formula: x i For training samples, y iFor category labels, K( is the weighting coefficient) The kernel function is defined as follows: Step S15: The classification results are converted into vector boundaries, and the misjudged areas are manually checked in conjunction with image texture, tone and aspect features to obtain the snow area distribution results; Step S20: Based on the multi-year snow identification results, the area, spatial distribution range and altitude range of snow patches in each year are statistically analyzed to extract the upper limit height, lower limit height and interannual variation characteristics of snow distribution, which are used to characterize the stability and fluctuation characteristics of multi-year snow spatial distribution; Step S30: Based on the obtained multi-year snow patches, the snow patches are classified and statistically analyzed according to the patch area level, and the proportion and structural changes of snow patches of different area levels in each year are analyzed to characterize the spatial pattern characteristics of snow patches evolving from dispersion to concentration or vice versa; Step S40: For the snow patches identified in each year, the corresponding spatial distribution of snow thickness is calculated, and the water storage of snow at the patch and regional scale is further calculated in conjunction with the density constraint model; the snow thickness is calculated by physical parameters such as base shear stress, topographic slope, density and shape parameters, and the calculation formula is:
[0028] In the formula: H is the snow thickness, The average base shear stress, and Here, θ represents the density of the base ice and the density of the snow, respectively, and θ is the slope. Let a be the acceleration due to gravity, and b be coefficients; where a and b are determined based on measured data, i.e., a and b are calculated by measuring a set of snow thickness samples.
[0029]
[0030] In the formula: S is the area of the snow-covered region, C is the perimeter of the boundary, and ΔH is the difference in elevation between the midstream lines; a snow density constraint function is constructed in the region without profile data, treating density as a function of altitude, temperature, and snow age, to extend the snow water equivalent to the unobserved area. The calculation formula is:
[0031] In the formula: Z is altitude, T is near-ground temperature, and τ is snow cover age; specifically, a linear fitting relationship between the three parameters (altitude, near-ground temperature, and snow cover age) and snow density is established, and then the specific relationship is derived.
[0032] The formula for calculating reserves is:
[0033] Where: hi For the thickness of the i-th layer of snow, and These are the densities of snow and water, respectively.
[0034] Step S50: Based on the area, thickness and reserves of snow patches over many years, construct statistical relationships between snow area and thickness, and between thickness and reserves, for the estimation of snow thickness and reserves in high mountains.
[0035] The above description is not intended to limit the present invention in any way. Although the present invention has been disclosed through the above embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints, characterized in that, Includes the following steps: Step S10: Acquire remote sensing image data of the target area at different time phases over several consecutive years, and acquire corresponding digital elevation model data. Based on remote sensing indices, identify the snow cover area of each year's images to obtain multi-year snow spatial distribution results and its patch vector data; Step S20: Based on the multi-year snow cover identification results, perform statistical analysis on the area, spatial distribution range, and altitude range of snow patches for each year, extract the upper and lower limits of snow distribution and their interannual variation characteristics to characterize the stability and fluctuation characteristics of multi-year snow spatial distribution; Step S30: Based on the obtained multi-year snow patches, Snow patches are classified and statistically analyzed according to their area levels. The proportion and structural changes of snow patches of different area levels in each year are analyzed to characterize the spatial pattern characteristics of snow patches evolving from dispersion to concentration or vice versa. Step S40: For the snow patches identified in each year, the spatial distribution of snow thickness is calculated, and the water storage of snow at the patch and regional scale is further calculated by combining the density constraint model. Step S50: Based on the area, thickness and storage data of snow patches over multiple years, the statistical relationship between snow area and thickness, and between thickness and storage is constructed for the estimation of snow thickness and storage in high mountains.
2. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 1, characterized in that, The specific process of identifying snow cover area based on remote sensing indices for images from different years in step S10 is as follows: Step S11: Perform radiometric calibration, atmospheric correction, cropping, and topographic correction on multi-temporal remote sensing images of the target area; Step S12: Construct a normalized snow cover index based on green light and shortwave infrared bands; Step S13: Use the normalized snow cover index threshold as an initial screening condition and mark pixels that meet the threshold as suspected snow cover areas; Step S14: Perform supervised classification on suspected snow cover areas; Step S15: Convert the classification results into vector boundaries, and manually verify the misjudged areas by combining image texture, tone, and slope aspect features, thereby obtaining the snow cover area distribution results.
3. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 2, characterized in that, The calculation formula in step S12 is: In the formula: and These are the green light and shortwave infrared reflectance of the remote sensing image, respectively. This is the normalized snow cover index.
4. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 2, characterized in that, In step S14, a support vector machine classification method is used for supervised classification.
5. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 4, characterized in that, The discriminant function of the support vector machine classification method is: In the formula: x i For training samples, y i For category labels, K( is the weighting coefficient) ) is the kernel function.
6. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 1, characterized in that, The formula for calculating the spatial distribution of snow thickness in step S40 is as follows: In the formula: H is the snow thickness, The average base shear stress, and Here, θ represents the density of the base ice and the density of the snow, respectively, and θ is the slope. Let be the acceleration due to gravity, and a and b be coefficients.
7. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 1, characterized in that, The density constraint model in step S40 is: In the formula: Z is the altitude, T is the near-surface temperature, and τ is the snow cover age.
8. The method for estimating the thickness and reserves of high-altitude snow cover based on remote sensing and physical constraints according to claim 1, characterized in that, The formula for calculating the reserves in step S40 is as follows: Where: h i For the thickness of the i-th layer of snow, and These are the densities of snow and water, respectively.