Remote sensing monitoring method and device for surface albedo of high-altitude area

Abstract

The invention relates to the technical field of remote sensing monitoring and climate analysis, in particular to a high-altitude area surface albedo remote sensing monitoring method and device, and the method comprises the steps: firstly obtaining the spectrum, infrared and topographic data of the surface, carrying out the radiation correction and topographic feature extraction, and generating preprocessing data; thirdly, generating multi-dimensional albedo data by using a multi-level fusion technology and an albedo calculation model, and obtaining joint albedo data through frequency domain and time domain analysis; and finally, generating a high-precision albedo product through a self-adaptive deep learning model and a residual error correction mechanism, and applying the high-precision albedo product to climate monitoring, ecological assessment and disaster early warning of a high-altitude area. According to the method, the accuracy and temporal-spatial resolution of albedo data are effectively improved, and reliable data support is provided for environmental research of high-altitude regions.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing monitoring and climate analysis, and in particular to a method and device for remote sensing monitoring of surface albedo in high altitude areas. Background Art

[0002] Remote sensing monitoring of surface albedo in high altitude areas is of great significance for understanding and predicting climate change, ecological environment and natural disasters in high altitude areas. Albedo is the ability of the surface to reflect solar radiation. Its changes directly affect the surface energy balance and are one of the important factors driving atmospheric movement. High altitude areas, especially the Qinghai-Tibet Plateau, are known as the "Third Pole of the Earth". The impact of its albedo changes on the global climate system cannot be ignored. However, existing technologies face some deficiencies and defects in albedo monitoring in high altitude areas.

[0003] The existing technology (Chinese invention patent, publication number: CN113946936B, name: Adaptive iterative sulfur dioxide inversion method based on EMI hyperspectral satellite payload) usually uses a combination of atmospheric radiation transfer model and atmospheric extinction law to calculate atmospheric albedo. This method mainly relies on the simulation and iterative fitting of the absorption characteristics of multiple gases. However, due to the complex terrain and changeable meteorological conditions in high-altitude areas, these traditional models often have problems with insufficient accuracy during the processing process. In addition, the existing methods lack comprehensive analysis of multi-dimensional albedo data over a large range of spatial and temporal scales, making it difficult to capture the dynamic changes and periodic characteristics of albedo, resulting in results that often deviate from reality, limiting the accurate prediction of climate change and ecological assessment. Summary of the invention

[0004] In view of the many problems existing in the above-mentioned prior art, the present invention provides a method and device for remote sensing monitoring of surface albedo in high altitude areas. The present invention obtains spectral data, infrared data and terrain data in high altitude areas, generates multi-dimensional albedo data using multi-level fusion and albedo calculation models, and analyzes and optimizes the data through adaptive deep learning models and residual correction mechanisms. The high-precision albedo products finally generated can be widely used in climate monitoring, ecological assessment and disaster warning in high altitude areas, significantly improving the prediction accuracy and system response capabilities.

[0005] A method for remote sensing monitoring of surface albedo in high altitude areas comprises the following steps:

[0006] Acquire the spectral data, infrared data and terrain data of the ground surface, perform radiation correction and terrain feature extraction on the acquired data, and generate pre-processed data;

[0007] The pre-processed data is fused at multiple levels, and multi-dimensional albedo data is generated through the albedo calculation model. The multi-dimensional albedo data is further analyzed in the frequency domain and time domain to generate joint albedo data.

[0008] The joint albedo data is input into the adaptive deep learning model for analysis, and the final corrected albedo data is generated through a dynamic feedback adjustment mechanism and a self-correction mechanism;

[0009] The final corrected albedo data will be integrated to generate high-precision albedo products for the surface in high-altitude areas, and will be used in climate monitoring, ecological assessment and disaster warning in high-altitude areas.

[0010] Preferably, the multi-level fusion includes fusing the preprocessed spectral data, infrared data and terrain data layer by layer according to the spatial dimension, spectral dimension and time dimension, wherein the spatial dimension fusion takes into account the characteristics of the surface of high-altitude areas, and the spectral data, infrared data and terrain data of the surface area of ​​each high-altitude area are fused separately to generate fused data.

[0011] Preferably, the multi-dimensional albedo data generated by the albedo calculation model is obtained by the following calculation formula:

[0012]

[0013] Where Albedo is the albedo of the surface at high altitudes; R up is the upward radiation reflected by the surface at high altitudes; R down is the incident downward radiation; R up It is obtained by measuring the radiation intensity reflected by the surface at high altitudes, R down It is obtained by measuring the sum of direct radiation from the sun and diffuse radiation from the atmosphere.

[0014] Preferably, the frequency domain and time domain analysis uses Fourier transform to convert multi-dimensional albedo data from the time domain to the frequency domain to identify the long-term trend and periodic characteristics of the surface albedo in high altitude areas, and uses the sliding average method to smooth the converted surface albedo data in high altitude areas to capture short-term fluctuations in the surface albedo in high altitude areas.

[0015] Preferably, the adaptive deep learning model is modeled using a multi-layer neural network structure, including an input layer for receiving joint albedo data and performing preliminary feature extraction, a hidden layer for performing complex feature extraction and pattern recognition on the joint albedo data through a nonlinear activation function, and an output layer for generating predicted albedo data.

[0016] Preferably, the adaptive deep learning model dynamically adjusts weights and biases in the following manner:

[0017] During the model training process, high-altitude environmental parameters are dynamically introduced, and the weights and biases of the adaptive deep learning model are adjusted according to the high-altitude environmental parameters.

[0018] Preferably, the dynamic feedback adjustment mechanism includes real-time monitoring of changes in surface albedo data in high-altitude areas, and adjusting parameters of the albedo calculation model based on historical data and current environmental conditions in high-altitude areas, to ensure that the albedo calculation model can adapt to environmental changes in high-altitude areas and optimize the final albedo results.

[0019] Preferably, the self-correction mechanism includes using a residual network structure to correct the prediction error, specifically comprising the following steps:

[0020] Calculate the difference between the predicted albedo data and the actually measured surface albedo data of high altitude areas to generate residual data;

[0021] The residual data is input into the residual network structure, and the residual data is corrected through multi-layer nonlinear transformation to reduce the error;

[0022] The corrected residual data are superimposed with the predicted albedo data to generate the corrected surface albedo data for high altitude areas.

[0023] Preferably, the application of the high-precision albedo product includes using the high-precision albedo product in climate monitoring, ecological assessment and disaster warning systems in high-altitude areas to improve the simulation of climate change in high-altitude areas, the assessment of the health status of ecosystems and the prediction of potential disasters in high-altitude areas.

[0024] A device for implementing the high-altitude area surface albedo remote sensing monitoring method, comprising:

[0025] The multi-source data acquisition module is used to obtain spectral data, infrared data and terrain data of the ground surface, and to perform radiation correction and terrain feature extraction on the acquired data to generate pre-processed data;

[0026] The data fusion and albedo calculation module is used to perform multi-level fusion of pre-processed data, generate multi-dimensional albedo data through the albedo calculation model, and further perform frequency domain and time domain analysis on the multi-dimensional albedo data to generate joint albedo data;

[0027] Intelligent analysis and dynamic adjustment module, which is used to receive the joint albedo data, input it into the adaptive deep learning model for analysis, and generate the final corrected albedo data through dynamic feedback adjustment mechanism and self-correction mechanism;

[0028] The albedo product generation and application module is used to integrate the final corrected albedo data to generate high-precision albedo products for the surface of high-altitude areas, and apply the high-precision albedo products to climate monitoring, ecological assessment and disaster warning in high-altitude areas.

[0029] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0030] The present invention uses multi-level data fusion technology to achieve accurate modeling of albedo data under complex terrain and changeable meteorological conditions in high-altitude areas, effectively solving the accuracy problem of existing technologies in albedo monitoring in high-altitude areas;

[0031] The present invention realizes frequency domain and time domain analysis of multi-dimensional albedo data by combining the albedo calculation model with the adaptive deep learning model, so that the present invention can capture the dynamic changes and periodic characteristics of the albedo, and greatly improve the spatiotemporal resolution and prediction ability of the data;

[0032] The present invention realizes real-time correction of albedo prediction errors through the innovative application of dynamic feedback adjustment mechanism and residual correction mechanism, so that albedo monitoring in high-altitude areas can maintain high accuracy under different environmental conditions, ensuring the reliability and stability of the results. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the process of the present invention;

[0034] Figure 2 It is a schematic diagram of multi-level data fusion in the present invention;

[0035] Figure 3 This is a schematic diagram of the structure of the adaptive deep learning model in the present invention;

[0036] Figure 4 Schematic diagram of the residual correction mechanism in the present invention;

[0037] Figure 5 It is a structural block diagram of the device of the present invention. DETAILED DESCRIPTION

[0038] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0039] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0040] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0041] like Figure 1 As shown, a method for remote sensing monitoring of surface albedo in high altitude areas comprises the following steps:

[0042] Acquire the spectral data, infrared data and terrain data of the ground surface, perform radiation correction and terrain feature extraction on the acquired data, and generate pre-processed data;

[0043] The purpose of acquiring spectral data, infrared data, and topographic data of the earth's surface is to comprehensively describe the physical and geometric characteristics of the earth's surface. Spectral data are acquired through multispectral or hyperspectral imaging technology, covering multiple bands from visible light to near infrared. These data can reflect the spectral characteristics of different materials on the earth's surface, such as vegetation, bare soil, snow, etc., which have significant differences in reflectivity in different bands. Infrared data are mainly acquired through thermal infrared imaging technology to capture the temperature distribution of the earth's surface. Especially in high-altitude areas, the thermal radiation characteristics of snow and glaciers are particularly important. These data are crucial for analyzing the energy balance of the earth's surface. Topographic data are acquired through LiDAR (Light Detection and Ranging) technology, which uses laser scanning to accurately measure the elevation changes of the earth's surface, thereby constructing a high-precision terrain model, which is particularly suitable for the refined depiction of complex terrain in high-altitude areas.

[0044] After obtaining these data, radiation correction and terrain feature extraction must be performed. Radiation correction is to eliminate the influence of the atmosphere on spectral and infrared data so that the data can truly reflect the characteristics of the surface. In high-altitude areas, radiation correction is particularly important due to the thin atmosphere and high solar radiation intensity. Specifically, the radiation correction process requires the use of an atmospheric radiation transfer model, combined with real-time atmospheric parameters (such as aerosol content, water vapor content, etc.) for inversion, to correct the atmospheric influence in spectral and infrared data, generate corrected data, and ensure the accuracy of these data.

[0045] Terrain feature extraction is to extract key terrain information from the acquired LiDAR data. Due to the complex terrain in high-altitude areas, such as mountains, canyons, and glacier areas, terrain feature extraction can help further understand the reflective characteristics of the surface. For example, the slope and azimuth of the surface will affect the incident angle of solar radiation, which in turn affects the measurement of albedo. Therefore, accurately extracting these terrain features can provide essential basic data for subsequent albedo calculations.

[0046] The pre-processed data is fused at multiple levels, and multi-dimensional albedo data is generated through the albedo calculation model. The multi-dimensional albedo data is further analyzed in the frequency domain and time domain to generate joint albedo data.

[0047] like Figure 2 As shown, preferably, the multi-level fusion includes fusing the preprocessed spectral data, infrared data and terrain data layer by layer according to the spatial dimension, spectral dimension and time dimension, wherein the spatial dimension fusion takes into account the characteristics of the surface of high-altitude areas, and the spectral data, infrared data and terrain data of the surface area of ​​each high-altitude area are fused separately to generate fused data.

[0048] Spatial dimension fusion is the basis of multi-level fusion. The surface features in high-altitude areas are extremely complex, with large terrain undulations and a large number of mountains, canyons and glaciers. These terrain features have a significant impact on the incident angle and intensity of solar radiation, which in turn affects the reflective characteristics of the surface. In spatial dimension fusion, these terrain features need to be fully considered, and the spectral data, infrared data and terrain data of each surface area need to be finely fused. For example, for a high-altitude mountainous area, the surfaces of different slopes and orientations have significant differences in spectral reflectance and thermal radiation, so these differences need to be fused into a unified data framework in the spatial dimension. This fusion not only retains the unique terrain features of each region, but also improves the consistency and accuracy of the overall data.

[0049] Next is spectral dimension fusion, which is mainly used to process data from different bands to improve the spectral accuracy of albedo. In high-altitude areas, since the spectral reflectance characteristics are greatly affected by the surface cover type (such as snow, vegetation, and bare soil) and atmospheric conditions, data from a single band often cannot fully reflect the spectral characteristics of the surface. Therefore, through spectral dimension fusion, the spectral data of different bands are combined with infrared data to more accurately reflect the reflective characteristics of surface materials. For example, the reflectivity of snow-covered areas in the visible light band and the near-infrared band is quite different. By fusing data from different bands, the scope and characteristics of snow cover can be more accurately identified. This fusion can significantly improve the reliability of spectral data and provide more accurate input for subsequent albedo calculations.

[0050] Time dimension fusion is the last dimension of multi-level fusion, which is mainly used to process data at different time points to capture the dynamic changes of surface albedo. The surface environment in high-altitude areas has a large time variability. For example, the seasonal melting of snow and the long-term retreat of glaciers will affect the albedo of the surface. In the time dimension fusion, these time-varying information can be integrated into the fused data by performing sequence analysis on data from different periods. For example, through the analysis of time series, the seasonal change pattern of snow cover can be identified and fused with the corresponding spectral and infrared data. This spatiotemporal fusion can help monitor and analyze the long-term change trends in high-altitude areas and provide important data support for predicting and responding to climate change.

[0051] Through the fusion of the above three dimensions, the fused data finally generated not only takes into account the terrain characteristics in space, enhances the complementarity between bands in the spectrum, but also captures the characteristics of dynamic changes in time. This multi-dimensional fusion method greatly improves the comprehensiveness and accuracy of the data, so that the remote sensing monitoring of surface albedo in high-altitude areas can more accurately reflect the actual surface conditions. Especially in applications such as climate monitoring, ecological assessment and disaster warning, this refined data fusion can significantly improve the prediction accuracy and stability of the model, thereby providing more reliable data support for decision-making.

[0052] Preferably, the multi-dimensional albedo data generated by the albedo calculation model is obtained by the following calculation formula:

[0053]

[0054] Where Albedo is the albedo of the surface at high altitudes; R up is the upward radiation reflected by the surface at high altitudes; R down is the incident downward radiation; R up It is obtained by measuring the radiation intensity reflected by the surface at high altitudes, R down It is obtained by measuring the sum of direct radiation from the sun and diffuse radiation from the atmosphere.

[0055] In high altitude areas, due to the differences in air pressure and temperature caused by the altitude, the energy balance of the earth's surface is significantly different from that of plain areas. Solar radiation, as the main source of energy on the earth's surface, will be partially absorbed and partially reflected by the earth's surface. The core of albedo calculation lies in accurately measuring and comparing the radiation reflected by the earth's surface (upward radiation) with the incident solar radiation (downward radiation), so as to obtain the specific value of the earth's surface reflection capacity.

[0056] R upRepresents the upward radiation reflected by the surface at high altitudes, specifically obtained by measuring the intensity of radiation reflected from the surface. This process usually involves using ground-based radiometers or satellite remote sensing equipment to record the radiant energy reflected from the surface into the atmosphere. This measurement is particularly important in high altitude areas because the radiation reflected from the surface can be affected by many factors such as snow cover and glacier surfaces, resulting in significant changes in albedo. For example, on the surface of a glacier, due to the extremely high reflectivity, R up The value will be significantly higher on surfaces covered with vegetation or exposed rock.

[0057] R down It represents the incident downward radiation, which is obtained by measuring the sum of direct radiation from the sun and diffuse radiation from the atmosphere. At high altitudes, due to the thin atmosphere and sparse clouds, the attenuation of solar radiation in the atmosphere is less, resulting in a relatively high intensity of downward radiation received by the surface. down The effects of scattering and absorption in the atmosphere need to be taken into account and are usually measured using a solar radiometer or calibrated remote sensing data.

[0058] The Albedo value calculated by the above formula reflects the energy reflection characteristics of the surface in high altitude areas. Due to the special environment of high altitude areas, the change of surface albedo depends not only on the surface type, but also on factors such as seasonal snow accumulation and glacier melting. Therefore, accurate calculation of albedo is crucial to understanding climate change, energy balance and ecosystem dynamics in high altitude areas.

[0059] Effect: The application of this albedo calculation model in remote sensing monitoring of surface albedo in high altitude areas can effectively improve the accuracy of analyzing the surface reflectivity. up and R down The albedo data generated can provide an important reference for climate monitoring, ecological assessment and disaster warning in high-altitude areas. For example, in climate models, albedo is one of the key parameters for calculating the surface energy balance, and its changes will directly affect the model's ability to predict climate factors such as temperature and precipitation. Especially in glacier and snow cover areas, the dynamic changes in albedo can be used as an indicator of important ecological processes such as glacier retreat and snow melt, thereby providing data support for ecological protection and management decisions.

[0060] Preferably, the frequency domain and time domain analysis uses Fourier transform to convert multi-dimensional albedo data from the time domain to the frequency domain to identify the long-term trend and periodic characteristics of the surface albedo in high altitude areas, and uses the sliding average method to smooth the converted surface albedo data in high altitude areas to capture short-term fluctuations in the surface albedo in high altitude areas.

[0061] As a commonly used signal processing tool, Fourier transform can convert time domain data into frequency domain to reveal the periodic characteristics and frequency components of the data. At high altitudes, the surface albedo not only fluctuates with seasonal changes, but is also affected by short-term climate events (such as blizzards and heat waves) and long-term geological changes (such as glacier retreat). Through Fourier transform, these complex time domain changes can be decomposed into components of different frequencies, thereby identifying long-term trends and periodic characteristics hidden in the time domain data. For example, through frequency domain analysis, the annual periodic changes in albedo can be identified, and the melting and accumulation patterns of seasonal snow can be inferred.

[0062] In the process of applying Fourier transform, the multi-dimensional albedo data must first be converted from the time domain to the frequency domain. In this step, the time series corresponding to each albedo data point will be converted into several frequency components, and the amplitude and phase information of each frequency component reveals the changing characteristics of the albedo within a specific period. This is especially important for high-altitude areas because the albedo in these areas is affected by a variety of periodic factors, such as seasonal snow cover, periodic meteorological changes, etc. By identifying these periodic changes, the long-term dynamics of the albedo can be better understood.

[0063] After the frequency domain analysis, the sliding average method is used to smooth the converted frequency domain data to capture the short-term fluctuations of the surface albedo. The sliding average method is a time series analysis tool that can effectively filter out random noise and retain the short-term trend of albedo by averaging the data within a certain range. In high-altitude areas, due to the drastic changes in the surface environment, the albedo may fluctuate significantly in the short term, such as avalanches, sudden changes in the surface after blizzards, etc. The sliding average method can smooth these fluctuations and highlight the true short-term change pattern.

[0064] By combining the Fourier transform and the sliding average method, frequency domain and time domain analysis can provide comprehensive insights into the surface albedo in high altitude areas. On the one hand, the Fourier transform reveals the periodic characteristics and long-term trends of the albedo, which is of great significance for understanding climate patterns and surface changes in high altitude areas. For example, analyzing annual periodic changes can help predict the time nodes of snow melt, thereby providing key data for water resource management and climate models. On the other hand, the application of the sliding average method ensures the accurate capture of short-term fluctuations in albedo, which is particularly important when responding to sudden climate events and conducting disaster warnings. For example, by capturing abnormal fluctuations in albedo before a landslide, early warnings can be issued to reduce potential disaster risks.

[0065] In the remote sensing monitoring of surface albedo in high altitude areas, frequency domain and time domain analysis can accurately identify the long-term trend, periodic characteristics and short-term fluctuations of albedo through the application of Fourier transform and sliding average method.

[0066] The joint albedo data is input into the adaptive deep learning model for analysis, and the final corrected albedo data is generated through a dynamic feedback adjustment mechanism and a self-correction mechanism;

[0067] like Figure 3 As shown, preferably, the adaptive deep learning model is modeled using a multi-layer neural network structure, including an input layer for receiving joint albedo data and performing preliminary feature extraction, a hidden layer for performing complex feature extraction and pattern recognition on the joint albedo data through a nonlinear activation function, and an output layer for generating predicted albedo data.

[0068] The adaptive deep learning model uses a multi-layer neural network structure to process and analyze complex joint albedo data. The core of the model is to extract useful features from a large amount of input data through recursive operations and nonlinear transformations of the multi-layer structure, and ultimately generate accurate prediction results.

[0069] At the input layer of the model, the joint albedo data is used as input. These data usually include the fusion results of spectral data, infrared data and terrain data. After preprocessing and fusion, these multi-source data still contain complex surface characteristics information, which must be initially extracted through the input layer of the model. At the input layer, the data is standardized or normalized to ensure that the numerical scales between different features are consistent, thereby improving the stability of the model and computational efficiency.

[0070] The hidden layer is the core of the model. It contains multiple neural network layers, each of which performs complex feature extraction and pattern recognition on the input data through nonlinear activation functions. Nonlinear activation functions (such as ReLU, Sigmoid, etc.) can introduce nonlinear factors, allowing the model to capture complex relationships in high-dimensional data. This is especially important for high-altitude areas because the surface albedo is affected by a variety of nonlinear factors, such as terrain undulations, climate change, and the complexity of surface cover types. Through the recursive operation of multi-layer neural networks, the model extracts deep-level features from the data layer by layer and gradually enhances the understanding of the surface albedo. For example, in glacier-covered areas, the hidden layer can identify the unique reflection pattern of the glacier surface and distinguish it from data in other areas, thereby improving the model's ability to predict glacier albedo.

[0071] The output layer is responsible for generating the final predicted albedo data. After multiple feature extractions and pattern recognitions in the hidden layer, the model is able to derive an accurate albedo value at the output layer, which represents the model's optimal estimate of the current input data. In practical applications, the albedo data generated by the output layer can be directly used for climate monitoring and environmental assessment in high-altitude areas. These predicted data can not only reflect the current albedo conditions of the surface, but also provide accurate predictions at different time points and environmental conditions, helping researchers and decision makers to better understand and respond to environmental changes in high-altitude areas.

[0072] The application of adaptive deep learning models in remote sensing monitoring of surface albedo in high altitude areas has a significant effect. Through the structure of multi-layer neural networks, the model is able to process complex multi-source data, extract key albedo features, and generate high-precision prediction data. This prediction capability enables the model to adapt to the changing environmental conditions in high altitude areas, whether in snow-covered mountains or in exposed rocky areas, the model can provide consistent and reliable albedo predictions.

[0073] For example, in a practical application scenario, researchers used an adaptive deep learning model to predict the albedo of a high-altitude glacier area. By training on historical data, the model successfully identified the characteristics of the glacier surface and was able to provide accurate albedo estimates in different seasons and weather conditions. These estimated data help researchers better understand the rate of glacier melting and its impact on the surrounding ecosystem. In addition, the model's prediction results can also be used in early warning systems to identify abnormal changes in albedo and predict potential natural disasters such as avalanches or glacier landslides in advance.

[0074] Preferably, the adaptive deep learning model dynamically adjusts weights and biases in the following manner:

[0075] During the model training process, high-altitude environmental parameters are dynamically introduced, and the weights and biases of the adaptive deep learning model are adjusted according to the high-altitude environmental parameters.

[0076] The core of the adaptive deep learning model lies in its adaptability, that is, the model can automatically adjust its parameters according to environmental changes to maintain prediction accuracy. In high-altitude areas, environmental conditions usually change greatly, such as sudden rise and fall in temperature, significant difference in radiation intensity between day and night, seasonal changes in snow cover, etc. These changes will directly affect the albedo of the surface. Therefore, the model must have the ability to adjust dynamically so that it can still provide accurate predictions under different environmental conditions.

[0077] During the model training process, environmental parameters of high altitude areas are dynamically introduced. These parameters may include temperature, humidity, altitude, air pressure, solar radiation intensity, etc. These environmental parameters are associated with the surface albedo as additional input data to guide the model's learning process. Specifically, the weights and biases of the model are important parameters that affect the model output, and they determine how the model responds to the input data. By dynamically adjusting the weights and biases, the model can better adapt to the environmental characteristics of high altitude areas. For example, under low temperature conditions, snow cover may increase the albedo of the surface. At this time, the model needs to adjust its weights so that the albedo prediction is more inclined to reflect the impact of snow cover. Similarly, under conditions of low air pressure and high radiation intensity, the model also needs to correct the baseline value of the albedo by adjusting the bias.

[0078] This dynamic adjustment is achieved through optimization algorithms, including the gradient descent method and the Adam optimization algorithm. During the training process, the model calculates the gradients of the current weights and biases based on the input environmental parameters, and updates the parameters based on these gradients. Through continuous iteration, the model gradually converges to the optimal parameter combination, so that the model can accurately predict the surface albedo under different environmental conditions.

[0079] This method of dynamically adjusting weights and biases enables the adaptive deep learning model to perform well in remote sensing monitoring of surface albedo in high-altitude areas. First, the dynamic adjustment allows the model to maintain stable prediction capabilities in the face of extreme environmental changes. Whether it is facing sudden meteorological events or seasonal environmental changes, the model can adapt in time and provide accurate albedo estimates. This is particularly critical for climate monitoring and disaster warning in high-altitude areas, because these areas often experience drastic environmental fluctuations, and traditional static models may not be able to cope with these changes.

[0080] For example, in one practical application, an adaptive deep learning model was used to predict the albedo of a valley in a high-altitude area. The area is affected by complex terrain, and the radiation conditions and temperature change dramatically. By introducing real-time environmental parameters, the model can dynamically adjust its weights and biases to accurately predict the albedo changes every day. In a sudden meteorological change, the temperature dropped sharply, resulting in an increase in snow cover. The model reflected this change in a timely manner through dynamic adjustments, and the prediction results accurately reflected the actual situation. This high-precision prediction not only helps researchers better understand environmental changes, but also provides key data support for local ecological protection and disaster prevention.

[0081] Preferably, the dynamic feedback adjustment mechanism includes real-time monitoring of changes in surface albedo data in high-altitude areas, and adjusting parameters of the albedo calculation model based on historical data and current environmental conditions in high-altitude areas, to ensure that the albedo calculation model can adapt to environmental changes in high-altitude areas and optimize the final albedo results.

[0082] The surface albedo in high altitude areas is affected by many factors, such as seasonal changes, weather conditions, terrain characteristics, and human activities. Changes in these factors can lead to dynamic fluctuations in surface albedo. In order to enable the albedo calculation model to still provide accurate predictions in such a complex environment, the dynamic feedback adjustment mechanism works in the following way.

[0083] The system monitors the albedo data of the surface in high altitude areas in real time. The monitoring content includes the real-time albedo values ​​obtained by remote sensing equipment, the changing trends of these values, and the comparison with historical data. Through this real-time monitoring, the system can quickly capture abnormal changes in albedo. For example, a sudden snowstorm may quickly change the albedo of the surface. At this time, the system can immediately detect this change and record its impact on the surface albedo.

[0084] After capturing the real-time data changes, the system compares these data with the historical albedo data of high altitude areas. The historical data provides a background reference for the model, which can help the model understand whether the current changes are normal periodic fluctuations or abnormal extreme events. Based on this data, the system also considers the current environmental conditions (such as temperature, humidity, radiation intensity, etc.), because these conditions are directly related to albedo changes. By combining real-time data, historical data and environmental conditions, the system can understand the current albedo changes more comprehensively and adjust the model parameters accordingly.

[0085] These parameter adjustments mainly involve key factors such as weights and biases in the albedo calculation model. By dynamically adjusting these parameters, the model can better reflect the actual situation of the current surface. For example, under extreme weather conditions (such as sudden snowstorm coverage), the system may adjust the weight of snow reflectivity in the model, making the model more inclined to predict higher albedo values, which is closer to the actual situation.

[0086] The dynamic feedback adjustment mechanism significantly improves the accuracy and adaptability of remote sensing monitoring of surface albedo in high-altitude areas. Through real-time monitoring and dynamic adjustment, the albedo calculation model can always maintain high-precision prediction capabilities under different environmental conditions. Especially in the face of sudden climate events or seasonal changes, the system can respond quickly and adjust the prediction model to adapt to new environmental conditions, avoiding inaccurate predictions due to parameter solidification.

[0087] For example, in actual applications, when the system detects a sudden increase in albedo in a high-altitude area, combined with historical data analysis, it is found that this is caused by snow cover. The system then automatically adjusts the model parameters so that in subsequent albedo predictions, the weight of snow reflectivity is increased, thereby improving the accuracy of the prediction. This dynamic adjustment not only improves the response speed of the system, but also ensures the robustness and stability of the model in long-term operation.

[0088] like Figure 4 As shown, preferably, the self-correction mechanism includes using a residual network structure to correct the prediction error, specifically including the following steps:

[0089] Calculate the difference between the predicted albedo data and the actually measured surface albedo data of high altitude areas to generate residual data;

[0090] The residual data is input into the residual network structure, and the residual data is corrected through multi-layer nonlinear transformation to reduce the error;

[0091] The corrected residual data are superimposed with the predicted albedo data to generate the corrected surface albedo data for high altitude areas.

[0092] The core of the self-correction mechanism is to use the residual network structure to correct the prediction error. The residual network is a deep learning model that can perform complex corrections and optimizations on the input data through multiple layers of nonlinear transformations. Specifically, the self-correction mechanism includes the following key steps:

[0093] The difference between the predicted albedo data and the actual measured albedo data at high altitudes is calculated to generate residual data. The purpose of this step is to quantify the deviation between the model prediction and the actual situation. The surface albedo at high altitudes is affected by various factors, such as snow cover, terrain undulations, meteorological conditions, etc., which may cause a certain error between the model prediction and the actual measurement. By calculating the residual data, the system can identify the size and distribution of these errors.

[0094] The generated residual data is input into the residual network structure for processing. The residual network performs complex feature extraction and correction on the input residual data through multiple layers of nonlinear transformation. Specifically, each layer in the network processes the residual data to gradually reduce the impact of the error. The depth and structural design of the network layer enable it to capture small changes in the error data and make fine corrections. For example, in glacier-covered areas, the prediction error of snow reflectivity may be large. The residual network can gradually reduce this error through multiple iterations, thereby improving the prediction accuracy of the model.

[0095] The corrected residual data is superimposed with the original predicted albedo data to generate the final corrected surface albedo data for high altitude areas. This step ensures that the final output data is closer to the actual measurement results by compensating the corrected error to the original prediction. This correction method not only improves the accuracy of the data, but also enhances the adaptability and stability of the model under various complex environmental conditions.

[0096] The application of the self-correction mechanism in remote sensing monitoring of surface albedo in high-altitude areas has significantly improved the accuracy and reliability of the prediction model. By correcting the prediction error in real time, the system can quickly respond to dynamic changes in the environment and reduce the deviation between prediction and reality. This is particularly important in high-altitude areas because the surface characteristics and environmental conditions in these areas are extremely complex, and traditional static models have difficulty in fully capturing these changes.

[0097] For example, in an actual application scenario, researchers used this self-correction mechanism to predict and monitor the albedo of a high-altitude area. There were certain differences between the initial prediction results and the actual measured data, especially in areas where snow cover and bare land alternated. These differences were calculated to generate residual data and input into the residual network for correction. The corrected data was superimposed on the original prediction to generate corrected albedo data. This process significantly reduced the error, making the corrected albedo data closer to the actual situation. The research results show that the accuracy of the corrected data in predicting glacier melting rates, snow cover changes, etc. has been significantly improved, which can provide more accurate data support for climate change research and disaster warning.

[0098] The final corrected albedo data will be integrated to generate high-precision albedo products for the surface in high-altitude areas, and will be used in climate monitoring, ecological assessment and disaster warning in high-altitude areas.

[0099] Preferably, the application of the high-precision albedo product includes using the high-precision albedo product in climate monitoring, ecological assessment and disaster warning systems in high-altitude areas to improve the simulation of climate change in high-altitude areas, the assessment of the health status of ecosystems and the prediction of potential disasters in high-altitude areas.

[0100] The generation process of high-precision albedo products is based on multi-layer processing and fine correction of raw remote sensing data, including data collection, radiation correction, multi-level fusion, frequency domain and time domain analysis, prediction of adaptive deep learning models, error correction of self-correction mechanisms, etc. After these steps, the albedo data has extremely high accuracy and precision, and can truly reflect the complex and changeable surface environment in high-altitude areas.

[0101] The generated high-precision albedo products are first used for climate monitoring. High-altitude areas have an important impact on the global climate system due to their special climatic conditions. For example, the Qinghai-Tibet Plateau is known as the "Third Pole of the Earth", and its albedo changes have an important impact on atmospheric circulation, monsoon systems, and global climate change. By applying high-precision albedo products, researchers can more accurately simulate and predict climate change in high-altitude areas and improve the reliability and predictive ability of climate models. Specifically, these albedo data can be used to improve the simulation of phenomena such as seasonal snow melt and glacier retreat, and provide more accurate input data for global climate change research.

[0102] Secondly, the application of albedo products in ecological assessment is crucial. Ecosystems in high-altitude areas are often more fragile and vulnerable to climate change and human activities. As an important indicator of surface energy balance, albedo can reflect key ecological variables such as surface vegetation conditions, soil moisture, and snow cover. Through high-precision albedo products, ecologists can assess the ecological health of different regions and identify the degradation or recovery process of ecosystems. For example, changes in albedo can indicate an increase or decrease in vegetation cover, thereby helping to formulate ecological protection and restoration plans.

[0103] In addition, high-precision albedo products also have important applications in disaster warning systems. Common natural disasters in high-altitude areas include avalanches, debris flows, glacier landslides, etc. The occurrence of these disasters is often related to abnormal changes in surface albedo. By real-time monitoring and analysis of albedo data, the early warning system can identify potential disaster risks. For example, when the albedo of a certain area is monitored to suddenly decrease, it may indicate that the snow accumulation is unstable and the risk of avalanches is increased. In this way, the disaster warning system can issue warnings earlier, helping relevant departments to take timely response measures to reduce casualties and property losses.

[0104] By applying high-precision albedo products to climate monitoring, ecological assessment and disaster warning systems, the environmental monitoring and management capabilities of high-altitude areas have been significantly improved. First, in terms of climate monitoring, the improvement in the accuracy of albedo data has significantly enhanced the prediction accuracy of climate models in simulating monsoon intensity, precipitation patterns, etc., especially in the prediction of long-term climate change trends. Secondly, in ecological assessments, the application of albedo products provides a scientific basis for assessing the health status of ecosystems, helping governments and research institutions to formulate more accurate ecological protection policies. Finally, in the field of disaster warning, the application of albedo products improves the response speed and accuracy of the warning system, allowing the risks of natural disasters to be identified in advance, greatly reducing the negative impact of disasters.

[0105] For example, in a typical application scenario, the research team used high-precision albedo products to conduct detailed monitoring of snow melting on the Qinghai-Tibet Plateau and successfully predicted the rate of glacier retreat. These data not only provide important support for global climate change research, but also help local governments better manage water resources, formulate disaster prevention and mitigation measures, and protect the ecological environment. This achievement demonstrates the great potential and value of high-precision albedo products in scientific research and practical applications.

[0106] like Figure 5 As shown, a device for implementing the high-altitude area surface albedo remote sensing monitoring method comprises:

[0107] The multi-source data acquisition module is used to obtain the spectral data, infrared data and terrain data of the surface, and perform radiation correction and terrain feature extraction on the acquired data to generate pre-processed data; through the multi-source data acquisition module, the system obtains the spectral data, infrared data and terrain data of the surface. After radiation correction and terrain feature extraction, these data generate pre-processed data to ensure the accuracy of subsequent analysis. This step uses high-precision sensor technology combined with a radiation correction model to eliminate atmospheric interference and other error sources, so that the data can more truly reflect the actual state of the surface.

[0108] The data fusion and albedo calculation module is used to perform multi-level fusion of pre-processed data, generate multi-dimensional albedo data through the albedo calculation model, and further perform frequency domain and time domain analysis on the multi-dimensional albedo data to generate joint albedo data; the data fusion and albedo calculation module performs multi-level fusion on these pre-processed data. The fusion process includes comprehensive processing of spatial, spectral and temporal dimensions to generate multi-dimensional albedo data. Subsequently, the system performs frequency domain and time domain analysis on these data through the albedo calculation model to generate joint albedo data. This process effectively combines traditional data processing methods with advanced algorithms, and can capture the dynamic changes and periodic characteristics of surface albedo, thereby improving the spatiotemporal resolution of monitoring.

[0109] The intelligent analysis and dynamic adjustment module is used to receive the joint albedo data, input it into the adaptive deep learning model for analysis, and generate the final corrected albedo data through the dynamic feedback adjustment mechanism and the self-correction mechanism; the intelligent analysis and dynamic adjustment module further analyzes the joint albedo data through the adaptive deep learning model. Through the dynamic feedback adjustment mechanism and the self-correction mechanism, the system can correct the model prediction error in real time and generate the final corrected albedo data. The key to this stage is that the system can adaptively adjust the model parameters to adapt to changes under different environmental conditions, thereby ensuring the high accuracy and reliability of the albedo prediction.

[0110] The albedo product generation and application module is used to integrate the final corrected albedo data to generate high-precision albedo products for the surface of high-altitude areas, and apply the high-precision albedo products to climate monitoring, ecological assessment and disaster warning in high-altitude areas. The albedo product generation and application module integrates these corrected albedo data to generate high-precision albedo products. These products are widely used in climate monitoring, ecological assessment and disaster warning in high-altitude areas, significantly improving the simulation accuracy of climate change, the accuracy of ecosystem health assessment, and the ability to predict potential natural disasters. Through this system, the complex surface albedo changes in high-altitude areas have been more accurately analyzed, providing strong data support for scientific research and practical applications.

[0111] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0112] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A remote sensing monitoring method for surface albedo in high altitude areas, characterized in that: The following steps are involved: Acquire the spectral data, infrared data and terrain data of the ground surface, perform radiation correction and terrain feature extraction on the acquired data, and generate pre-processed data; The pre-processed data is fused at multiple levels, and multi-dimensional albedo data is generated through the albedo calculation model. The multi-dimensional albedo data is further analyzed in the frequency domain and time domain to generate joint albedo data. The joint albedo data is input into the adaptive deep learning model for analysis, and the final corrected albedo data is generated through a dynamic feedback adjustment mechanism and a self-correction mechanism; The final corrected albedo data will be integrated to generate high-precision albedo products for the surface in high-altitude areas, and will be used in climate monitoring, ecological assessment and disaster warning in high-altitude areas.

2. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The multi-level fusion includes fusing the pre-processed spectral data, infrared data and terrain data layer by layer according to the spatial dimension, spectral dimension and time dimension, wherein the spatial dimension fusion takes into account the surface characteristics of high-altitude areas, and fuses the spectral data, infrared data and terrain data of the surface area of ​​each high-altitude area respectively to generate fused data.

3. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The multi-dimensional albedo data generated by the albedo calculation model is obtained by the following calculation formula: Where Albedo is the albedo of the surface at high altitudes; R up is the upward radiation reflected by the surface at high altitudes; R down is the incident downward radiation; R up It is obtained by measuring the radiation intensity reflected by the surface at high altitudes, R down It is obtained by measuring the sum of direct radiation from the sun and diffuse radiation from the atmosphere.

4. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The frequency domain and time domain analysis uses Fourier transform to convert multi-dimensional albedo data from the time domain to the frequency domain to identify the long-term trend and periodic characteristics of the surface albedo in high altitude areas, and uses the sliding average method to smooth the converted high altitude surface albedo data to capture the short-term fluctuations of the surface albedo in high altitude areas.

5. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The adaptive deep learning model is modeled using a multi-layer neural network structure, including an input layer for receiving joint albedo data and performing preliminary feature extraction, a hidden layer for performing complex feature extraction and pattern recognition on the joint albedo data through a nonlinear activation function, and an output layer for generating predicted albedo data.

6. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The adaptive deep learning model dynamically adjusts weights and biases in the following ways: During the model training process, high-altitude environmental parameters are dynamically introduced, and the weights and biases of the adaptive deep learning model are adjusted according to the high-altitude environmental parameters.

7. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The dynamic feedback adjustment mechanism includes real-time monitoring of changes in surface albedo data in high-altitude areas, and adjusting parameters of the albedo calculation model based on historical data and current environmental conditions in high-altitude areas, to ensure that the albedo calculation model can adapt to environmental changes in high-altitude areas and optimize the final albedo results.

8. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The self-correction mechanism includes using a residual network structure to correct the prediction error, specifically including the following steps: Calculate the difference between the predicted albedo data and the actually measured surface albedo data of high altitude areas to generate residual data; The residual data is input into the residual network structure, and the residual data is corrected through multi-layer nonlinear transformation to reduce the error; The corrected residual data are superimposed with the predicted albedo data to generate the corrected surface albedo data for high altitude areas.

9. The method for remote sensing monitoring of surface albedo in high altitude areas according to claim 1, characterized in that: The application of the high-precision albedo product includes using the high-precision albedo product in climate monitoring, ecological assessment and disaster warning systems in high-altitude areas to improve the simulation of climate change in high-altitude areas, the assessment of the health status of ecosystems and the prediction of potential disasters in high-altitude areas.

10. A device for implementing the remote sensing monitoring method of surface albedo in high altitude areas according to any one of claims 1 to 9, characterized in that: include: The multi-source data acquisition module is used to obtain spectral data, infrared data and terrain data of the ground surface, and to perform radiation correction and terrain feature extraction on the acquired data to generate pre-processed data; The data fusion and albedo calculation module is used to perform multi-level fusion of pre-processed data, generate multi-dimensional albedo data through the albedo calculation model, and further perform frequency domain and time domain analysis on the multi-dimensional albedo data to generate joint albedo data; Intelligent analysis and dynamic adjustment module, which is used to receive the joint albedo data, input it into the adaptive deep learning model for analysis, and generate the final corrected albedo data through dynamic feedback adjustment mechanism and self-correction mechanism; The albedo product generation and application module is used to integrate the final corrected albedo data to generate high-precision albedo products for the surface of high-altitude areas, and apply the high-precision albedo products to climate monitoring, ecological assessment and disaster warning in high-altitude areas.

Images