Ecological sensitivity evaluation method for arid region

Abstract

The invention relates to the technical field of arid region ecological evaluation, and discloses an arid region ecological sensitivity evaluation method, which comprises the following steps: S1, multivariate data acquisition: acquiring land and land data by utilizing a satellite remote sensing image, acquiring topographic data by utilizing a DEM (Digital Elevation Model), acquiring a spectrum at a mineral sensitive wave band by utilizing a hyperspectral sensor, and acquiring a mineral sensitive wave band; a low-altitude unmanned aerial vehicle is used for carrying a multispectral camera to obtain vegetation disease and pest data, and the data are preprocessed and fused; s2, factor selection: selecting a landform factor, a vegetation factor, a soil factor, a water resource factor, a climate factor, a human activity factor, an ecological system service function value factor and a landscape connectivity factor; and S3, multi-factor grading evaluation: dividing the factor original data into five grades. Through multivariate data acquisition, hyperspectral remote sensing is used for acquiring mineral spectrums, and the unmanned aerial vehicle is used for acquiring fine vegetation disease and insect pest data, so that the problem of unreliable evaluation results caused by single source of existing ecological sensitivity evaluation data is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological evaluation in arid regions, and specifically to a method for evaluating ecological sensitivity in arid regions. Background Technique

[0002] The ecological system in arid regions is special and fragile, occupying a considerable proportion of the earth's land area. Its remarkable feature is arid climate, scarce and variable precipitation, with the annual precipitation often less than 250 millimeters, while the evaporation rate is much higher than the precipitation, which makes water the key factor restricting the development of the ecological system. In terms of vegetation, it is mainly composed of drought-tolerant and sand-resistant plants, such as cacti and alhagi sparsifolia. These plants have evolved special structures. The fleshy stems of cacti can store water, and the root systems of alhagi sparsifolia are extremely developed, which can penetrate more than ten meters underground to absorb water, so as to adapt to the harsh environment. Animals also mostly have water-saving characteristics. For example, the ears of fennec foxes are large, which is conducive to heat dissipation and reducing water loss; camels can go without drinking water for a long time and rely on humps to store fat to maintain metabolism. The soil generally has low fertility, loose texture, and is severely eroded by wind, easily forming mobile sand dunes. Where water sources are relatively abundant, oasis ecosystems will appear, which are the "shelters" of life in arid regions. Nourished by mountain snowmelt or groundwater, they give birth to relatively diverse biological communities and develop irrigation agriculture. However, generally speaking, the ecological system in arid regions has poor stability. Unreasonable human activities such as over-reclamation and over-grazing are extremely likely to break the ecological balance and cause ecological problems such as land desertification and biodiversity reduction.

[0003] The existing data sources for ecological sensitivity evaluation are single and the accuracy is insufficient, making it difficult to comprehensively reflect the complex ecological conditions in arid regions, thus resulting in unreliable evaluation results. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technology, the present invention provides a method for evaluating ecological sensitivity in arid regions, which solves the problems that the existing data sources for ecological sensitivity evaluation are single and the accuracy is insufficient, making it difficult to comprehensively reflect the complex ecological conditions in arid regions, thus resulting in unreliable evaluation results.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for evaluating ecological sensitivity in arid regions includes the following steps:

[0006] S1. Acquisition of multi-source data: Collect land data using satellite remote sensing images, obtain topographic data using DEM, collect spectra in mineral-sensitive bands using hyperspectral sensors, obtain vegetation pest and disease data using low-altitude drones equipped with multispectral cameras, and preprocess and fuse the data;

[0007] S2. Factor selection: Select topographic and geomorphic factors, vegetation factors, soil factors, water resource factors, climate factors, human activity factors, ecological system service function value factors, and landscape connectivity factors;

[0008] S3. Multi-factor hierarchical evaluation: The original factor data is divided into 5 levels. The ecosystem service function value factors are classified according to the model and the characteristics of the arid area, and the landscape connectivity factors are classified into three levels according to the index. GIS is used to transform the classification.

[0009] S4. Fuzzy weight determination: The analytic hierarchy process is adopted, and a model is constructed by integrating fuzzy mathematics. The expert judgment is converted into a fuzzy matrix, and the weight is calculated through verification.

[0010] S5. Accurate score calculation and classification: The weighted overlay method is used to calculate the comprehensive score, and a model is established in combination with the random forest algorithm, verified and calibrated, and sensitive sub-levels are refined and added.

[0011] S6. Comprehensive evaluation and application: GIS analysis is used to construct a model, and big data is used to mine the impact of human activities. Adaptive management of ecological restoration is introduced in the planning, and the strategy is adjusted according to the changes in sensitive areas.

[0012] Preferably, in the S1, the satellite remote sensing image adopts a multi-spectral image, and radiometric calibration and geometric correction are carried out using remote sensing image processing software; the DEM data is obtained by measuring with surveying instruments.

[0013] Preferably, in the S1, the working band of the hyperspectral sensor covers 1000 - 2500 nm, and the spectral resolution is below 10 nm; the flight altitude of the low-altitude unmanned aerial vehicle is maintained at 50 - 200 meters, and the multi-spectral camera has 4 - 8 bands, which is used to accurately obtain vegetation pest data. After the data is obtained, key information is screened through feature extraction algorithms for preprocessing.

[0014] Preferably, in the S2, the ecosystem service function value factors are modeled to quantify the impact, and the landscape connectivity factors analyze the connectivity through graph theory and circuit theory to measure ecological sensitivity; the modeling to quantify the impact combines the ecological characteristics of the arid area, and the water balance method and the universal soil loss equation method are used to quantify the water conservation and soil conservation functions; the analysis of the landscape connectivity factors uses Graphab software to calculate the effective connectivity index according to circuit theory.

[0015] Preferably, in the S3, dividing the original factor data into 5 levels includes extremely sensitive, highly sensitive, moderately sensitive, slightly sensitive, and insensitive. Using GIS to transform the classification can output vector and raster layers.

[0016] Preferably, in the S4, when constructing a model by the fuzzy analytic hierarchy process, triangular fuzzy numbers are used to represent expert judgment. The fuzzy consistency test calculates the fuzzy consistency index. When the index is less than 0.1, the judgment matrix is considered valid, and the eigenvector method is used to calculate the factor weights.

[0017] Preferably, in S5, the weighted superposition method calculates the comprehensive score according to the formula Score = ∑(Wi × Si), where Wi is the factor weight and Si is the factor sensitivity evaluation value; the random forest algorithm is implemented using the Scikit-learn library in Python, and the model is optimized by adjusting the number of decision trees and the maximum depth, and is compared and analyzed with the weighted superposition result.

[0018] Preferably, in S5, when refining the sensitivity level, the clustering analysis method is adopted. Based on the comprehensive score data distribution, 2-3 sub-levels are added in the highly sensitive and extremely sensitive areas to clarify the score intervals and ecological characteristics of each sub-level.

[0019] Preferably, in S6, the GIS software is used for ecological sensitivity analysis, combined with spatial autocorrelation analysis and buffer analysis to generate a special map of ecological sensitivity; big data mining uses the Hadoop platform to process social media and mobile device location data through the MapReduce programming model.

[0020] Preferably, in S6, the adaptive management of ecological restoration formulates restoration plans for 1-2 years, 3-5 years, and more than 5 years according to the ecological sensitivity evaluation results, monitors the ecological indicators of sensitive areas every six months, and dynamically adjusts the restoration strategy according to the monitoring results.

[0021] The present invention provides a method for evaluating ecological sensitivity in arid areas. It has the following beneficial effects:

[0022] 1. In the present invention, through the acquisition of multi-source data, hyperspectral remote sensing is used to obtain mineral spectra, drones are used to obtain fine vegetation pest data, combined with satellite remote sensing images and DEM, greatly enriching the data dimension, comprehensively grasping the ecological background information of arid areas, solving the problems of incomplete data and insufficient accuracy, providing a solid foundation for subsequent evaluation, and thus improving the problem that the existing data sources for ecological sensitivity evaluation are single, with insufficient accuracy, and it is difficult to comprehensively reflect the complex ecological conditions in arid areas, which will result in unreliable evaluation results.

[0023] 2. In the present invention, through the ecosystem service function value factor and the landscape connectivity factor, the evaluation system is improved from the perspective of the integrity of the ecosystem function and structure, making the evaluation result more in line with the ecological reality in arid areas and avoiding evaluation deviation caused by factor omission.

[0024] 3. In the present invention, the fuzzy analytic hierarchy process model is constructed by integrating fuzzy mathematics, making the determination of weights more scientific and reasonable; combining the random forest algorithm and the weighted superposition method to calculate the comprehensive score, mutually verifying and calibrating, refining the sensitivity level, and improving the accuracy and reliability of the evaluation, solving the problem that the evaluation method is not precise enough.

[0025] 4. In the present invention, through big data mining technology, integrating multi-source spatio-temporal data such as social media and mobile device positioning, the impact of human activities on ecological sensitivity can be deeply analyzed, providing a more comprehensive basis for ecological protection and planning, and making up for the deficiencies of traditional methods in human activity monitoring and analysis.

[0026] 5. In the present invention, ecological restoration adaptive management is introduced in the planning. According to the changes in sensitive areas, the strategy is adjusted in real time, short-term, medium-term, and long-term restoration plans are formulated and monitored regularly to ensure that the ecological restoration measures match the actual changes in the ecosystem, improving the ecological restoration effect and the stability of the ecosystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work fall within the scope of protection of the present invention.

[0029] Please refer to the attached Figure 1 , an embodiment of the present invention provides a method for evaluating ecological sensitivity in arid areas, which is characterized in that it includes the following steps:

[0030] S1. Multivariate data acquisition: Using satellite remote sensing images to collect land data, using DEM to obtain topographic data, using hyperspectral sensors to collect spectra in mineral-sensitive bands, using low-altitude unmanned aerial vehicles equipped with multispectral cameras to obtain vegetation pest data, and preprocessing and fusing the data;

[0031] S2. Factor selection: Selecting topographic and geomorphic factors, vegetation factors, soil factors, water resource factors, climate factors, human activity factors, ecological system service function value factors, and landscape connectivity factors;

[0032] S3. Multi-factor grading evaluation: Dividing the original factor data into 5 levels, grading the ecological system service function value factors according to the model and the characteristics of the arid area, grading the landscape connectivity factors into three levels according to the index, and using GIS to transform the grading;

[0033] S4. Fuzzy weight determination: Using the analytic hierarchy process, integrating fuzzy mathematics to construct a model, transforming expert judgments into a fuzzy matrix, and calculating the weights through verification;

[0034] S5. Accurate score calculation and grading: Calculating the comprehensive score using the weighted superposition method, establishing a model in combination with the random forest algorithm, verifying and calibrating, and refining and adding sensitive sub-levels;

[0035] S6, Comprehensive evaluation and application: Analyze using GIS, build models, borrow big data to mine the impacts of human activities, introduce adaptive management of ecological restoration in planning, and adjust strategies according to changes in sensitive areas.

[0036] In S1, the satellite remote sensing images use multispectral images, and radiometric calibration and geometric correction are carried out using remote sensing image processing software; the DEM data is obtained by measuring with surveying instruments.

[0037] Specifically, in arid regions, the mineral composition is closely related to the properties, fertility of the soil, and the vegetation growth environment. For example, the presence or absence of certain minerals may affect the water and fertilizer retention capacity of the soil, and thus affect the distribution and growth status of vegetation, which is one of the important factors in ecological sensitivity evaluation. In addition, the mineral spectral data may also reflect information such as the potential distribution of groundwater resources, etc., all of which play an important role in comprehensively evaluating the ecological sensitivity of arid regions. By using multispectral satellite remote sensing images and using professional software for radiometric calibration and geometric correction, it can ensure accurate radiometric information and precise spatial position of the images; through radiometric calibration, the pixel values of the images correspond to the true radiance of the ground objects, laying a foundation for quantitatively analyzing the spectral characteristics of ground objects, which is conducive to subsequent accurate identification and differentiation of different ground object types, such as vegetation, water bodies, bare land, etc.; through geometric correction, the geometric distortions caused by sensors, satellite attitudes, and the earth's curvature are eliminated, making the images conform to the map projection coordinate system, facilitating the splicing of multi-scene images and the overlay analysis with other geographical data; by measuring with professional surveying instruments to obtain DEM data, it can accurately reflect the ground elevation information, combined with satellite remote sensing images, it can intuitively present the undulation of the terrain and assist in analyzing the impact of terrain on ecological factors, such as the effects of slope and aspect on the redistribution of light, heat, water, and vegetation distribution, providing reliable terrain basic data for comprehensively evaluating the ecological sensitivity of arid regions.

[0038] In S1, the working band of the hyperspectral sensor covers 1000 - 2500 nm, and the spectral resolution is below 10 nm; the flight altitude of the low-altitude unmanned aerial vehicle remains at 50 - 200 meters, and the multispectral camera has 4 - 8 bands, which are used to accurately obtain vegetation pest data, and after the data is obtained, key information is screened through feature extraction algorithms for preprocessing.

[0039] Specifically, the working band of the hyperspectral sensor covers 1000 - 2500 nm and the spectral resolution is below 10 nm. It can obtain rich and detailed spectral feature information of ground objects. This band range is sensitive to soil minerals, vegetation biochemical components, etc., and can accurately identify different mineral types. By analyzing the changes in spectral features, it can monitor the health status of vegetation, helping to detect early signs of pests and diseases. With a low-altitude drone flying at an altitude of 50 - 200 meters and carrying a multi-spectral camera with 4 - 8 bands, it can obtain vegetation images with high spatial resolution. The low-altitude flight ensures that the obtained images can clearly present the details of vegetation. The multi-spectral bands are designed according to the reflection and absorption characteristics of vegetation pests and diseases, facilitating the accurate identification of the occurrence area and degree of pests and diseases. After data acquisition, key information is screened through feature extraction algorithms, which can remove redundant data and retain core features such as spectra and textures related to pests and diseases, providing efficient and accurate data support for subsequent data preprocessing and pest and disease analysis, and improving the accuracy of monitoring and assessment of vegetation pests and diseases in arid areas.

[0040] In S2, the value factors of ecosystem service functions are modeled to quantify the impacts. The landscape connectivity factor analyzes the connectivity through graph theory and circuit theory to measure ecological sensitivity. For the modeling and quantification of impacts, combined with the ecological characteristics of arid areas, the water balance method and the universal soil loss equation method are used to quantify the functions of water conservation and soil conservation. The analysis of the landscape connectivity factor uses Graphab software to calculate the effective connectivity index according to circuit theory.

[0041] Specifically, by modeling and quantifying the impacts of the value factors of ecosystem service functions, and combining with the ecological characteristics of arid areas, the water balance method and the universal soil loss equation are used to quantify the functions of water conservation and soil conservation, which can accurately evaluate the specific value of ecosystem service functions. This helps to clarify the contributions of various elements of the ecosystem to the ecological environment in arid areas, providing a scientific basis for the rational allocation of resources and ecological compensation. The landscape connectivity factor is analyzed through graph theory and circuit theory, and the Graphab software is used to calculate the effective connectivity index, which can measure the connectivity between landscape patches. From this, it can reveal the structural integrity of the ecosystem and the ease of biological migration and diffusion, and identify key ecological corridors and fragmented areas. Combining the quantification results of the value of ecosystem service functions can comprehensively and integrally measure ecological sensitivity, providing a more accurate direction for the formulation of ecological protection plans and restoration strategies.

[0042] In S3, the original factor data is divided into 5 levels, including extremely sensitive, highly sensitive, moderately sensitive, slightly sensitive, and insensitive. Using GIS to transform the classification can output vector and raster layers.

[0043] Specifically, the original factor data are divided into five levels and converted into vector and raster layers using GIS, which can intuitively present the spatial sensitivity distribution of each ecological factor; the vector layer can express the factor boundary and location information, facilitating the analysis of the ecological sensitivity characteristics of a specific area; the raster layer can display continuous spatial changes and reflect the gradual trend of ecological sensitivity; this helps to quickly identify ecologically sensitive areas, provide a visual basis for ecological protection planning, resource management, etc., support targeted decision-making, and rationally arrange protection measures and resource allocation.

[0044] In S4, when constructing the model using the fuzzy analytic hierarchy process, triangular fuzzy numbers are used to represent expert judgments. The fuzzy consistency test is performed by calculating the fuzzy consistency index. When the index is less than 0.1, the judgment matrix is considered valid, and the eigenvector method is used to calculate the factor weights.

[0045] Specifically, this can reduce the subjectivity of expert judgment, and triangular fuzzy numbers can more comprehensively express the uncertainty of expert opinions, making the judgment results more in line with the actual situation; the fuzzy consistency test ensures the rationality and coherence of the judgment matrix logic, and the matrix is considered valid when the index is less than 0.1, which enhances the reliability of weight calculation; the eigenvector method can accurately calculate the weight of each factor based on the valid matrix, making the weight distribution more scientific, thereby improving the accuracy and credibility of the ecological sensitivity assessment results.

[0046] In S5, the weighted superposition method calculates the comprehensive score according to the formula Score = ∑(Wi × Si), where Wi is the factor weight and Si is the factor sensitivity evaluation value; the random forest algorithm is implemented using Python's Scikit-learn library, and the model is optimized by adjusting the number of decision trees and the maximum depth, and then compared with the weighted superposition results.

[0047] Specifically, the comprehensive score is calculated by the weighted superposition method according to the formula Score = ∑(Wi×Si), which can integrate the weights of each factor and the sensitivity evaluation value to obtain an intuitive numerical value that reflects the overall ecological sensitivity, providing a quantitative basis for the preliminary assessment; the random forest algorithm, with the help of Python's Scikit-learn library, can adjust the number of decision trees and optimize the model with the maximum depth, so as to explore the complex potential relationships of the data and improve the prediction accuracy; by comparing and analyzing the results of the two, they can be verified and calibrated with each other; the results of the weighted superposition method can assist in understanding the prediction trend of the model, and the results of the random forest algorithm can correct the possible deviations of the weighted superposition method, thereby optimizing the comprehensive score of ecological sensitivity and improving the evaluation accuracy.

[0048] In S5, when refining the sensitivity level, the cluster analysis method is used. Based on the distribution of comprehensive score data, 2-3 sub-levels are added in highly sensitive and extremely sensitive areas, and the score range and ecological characteristics of each sub-level are clarified.

[0049] Specifically, by refining the sensitivity level through cluster analysis, potential patterns in the comprehensive score data can be mined. Adding sub-levels in the highly sensitive and extremely sensitive areas can more precisely distinguish the subtle differences in ecological sensitivity. At the same time, the score intervals and ecological characteristics of each sub-level can be clarified, providing more detailed information for ecological protection. Differentiated protection strategies can be formulated for different sub-levels, giving priority to protecting areas with more vulnerable ecological characteristics, enhancing the pertinence and effectiveness of ecological protection, and also helping to gain a deeper understanding of the state of the ecosystem under different sensitivity levels.

[0050] In S6, the GIS software is used for ecological sensitivity analysis, combined with spatial autocorrelation analysis and buffer analysis to generate a thematic map of ecological sensitivity. Big data mining uses the Hadoop platform to process social media and mobile device location data through the MapReduce programming model.

[0051] Specifically, by using the GIS software combined with spatial autocorrelation analysis and buffer analysis to generate a thematic map of ecological sensitivity, the spatial distribution characteristics and aggregation trends of ecological sensitivity can be visually presented. Through spatial autocorrelation analysis, the degree of spatial correlation of ecological sensitivity can be revealed, and buffer analysis can determine the influence range of different ecological elements. The thematic map provides clear spatial information for decision-makers to assist in formulating regional ecological protection plans. By using the Hadoop platform and the MapReduce programming model to process social media and mobile device location data, the correlation between human activities and ecological sensitivity can be efficiently mined. Social media data reflects human behavior and interests, and mobile device location data presents the population activity trajectory. Through big data analysis, human activity hotspots in areas with high ecological sensitivity can be identified, providing data support for evaluating the impact of human activities on the ecosystem and helping to fully consider human factors when formulating ecological protection strategies.

[0052] In S6, based on the results of the ecological sensitivity assessment, ecological restoration adaptive management formulates restoration plans for 1 - 2 years, 3 - 5 years, and more than 5 years, monitors the ecological indicators in the sensitive areas every six months, and dynamically adjusts the restoration strategy according to the monitoring results.

[0053] Specifically, through the formulation of restoration plans at different time scales, adaptive management of ecological restoration can comprehensively and orderly carry out ecological restoration work. Short-term plans of 1-2 years can quickly respond to urgent ecological problems, such as controlling local soil erosion; medium-term plans of 3-5 years can initially restore the structure of the ecosystem, such as the reconstruction of vegetation communities; long-term plans of more than 5 years can comprehensively restore and enhance the functions of the ecosystem. By monitoring ecological indicators in sensitive areas every six months, the dynamic changes of the ecosystem can be timely grasped. Adjusting restoration strategies dynamically based on the monitoring results can ensure that restoration measures always fit the actual situation of the ecosystem. When it is found that the vegetation restoration rate fails to meet the expectations, the planting plan can be adjusted in a timely manner. If it is monitored that soil erosion intensifies, soil and water conservation measures can be quickly strengthened, thereby improving the scientificity and effectiveness of ecological restoration and ensuring that the ecosystem gradually develops in a good direction.

[0054] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An ecological sensitivity evaluation method in arid regions, characterized in that: It includes the following steps: S1. Multivariate data acquisition: Using satellite remote sensing images to collect land data, using DEM to obtain topographic data, using hyperspectral sensors to collect spectra in mineral-sensitive bands, using low-altitude drones equipped with multispectral cameras to obtain vegetation pest and disease data, and preprocessing and fusing the data; S2. Factor selection: Selecting topographic and geomorphic factors, vegetation factors, soil factors, water resource factors, climate factors, human activity factors, ecosystem service function value factors, and landscape connectivity factors; S3. Multi-factor hierarchical evaluation: Classifying the original factor data into 5 levels, classifying the ecosystem service function value factors according to the model and the characteristics of arid areas, classifying the landscape connectivity factors into three levels according to the index, and using GIS to transform the classification; S4. Fuzzy weight determination: Using the analytic hierarchy process, integrating fuzzy mathematics to construct a model, transforming expert judgments into a fuzzy matrix, and calculating the weights through verification; S5. Accurate score calculation and classification: Calculating the comprehensive score using the weighted overlay method, establishing a model in combination with the random forest algorithm, verifying and calibrating, refining and adding sensitive sub-levels; S6. Comprehensive evaluation and application: Using GIS analysis, constructing a model, using big data to mine the impact of human activities, introducing ecological restoration adaptive management in planning, and adjusting strategies according to the changes in sensitive areas.

2. The ecological sensitivity evaluation method in arid regions according to claim 1, wherein: In S1, the satellite remote sensing images are multispectral images, and radiometric calibration and geometric correction are carried out using remote sensing image processing software; the DEM data is obtained by measuring with surveying instruments.

3. The ecological sensitivity evaluation method in arid regions according to claim 1, characterized in that: In S1, the working band of the hyperspectral sensor covers 1000 - 2500 nm, and the spectral resolution is below 10 nm; the flight altitude of the low-altitude drone is maintained at 50 - 200 meters, and the multispectral camera has 4 - 8 bands, which are used to accurately obtain vegetation pest and disease data. After the data is obtained, key information is screened through feature extraction algorithms for preprocessing.

4. The ecological sensitivity evaluation method for arid regions according to claim 1, characterized in that: In S2, the ecosystem service function value factors are modeled to quantify the impact, and the landscape connectivity factors analyze the connectivity through graph theory and circuit theory to measure ecological sensitivity; the modeling to quantify the impact combines the ecological characteristics of arid areas, and the water balance method and the universal soil loss equation method are used to quantify the water conservation and soil conservation functions; the analysis of the landscape connectivity factors uses Graphab software to calculate the effective connectivity index based on circuit theory.

5. The ecological sensitivity evaluation method in an arid area according to claim 1, characterized in that: In S3, classifying the original factor data into 5 levels includes extremely sensitive, highly sensitive, moderately sensitive, slightly sensitive, and insensitive. Using GIS to transform the classification can output vector and raster layers.

6. The ecological sensitivity evaluation method for arid regions according to claim 1, characterized in that: In S4, when constructing a model using the fuzzy analytic hierarchy process, triangular fuzzy numbers are used to represent expert judgments. The fuzzy consistency test calculates the fuzzy consistency index. When the index is less than 0.1, the judgment matrix is considered valid, and the eigenvector method is used to calculate the factor weights.

7. The ecological sensitivity evaluation method in arid regions according to claim 1, characterized in that: In S5, the weighted superposition method calculates the comprehensive score according to the formula Score = ∑(Wi × Si), where Wi is the factor weight and Si is the factor sensitivity evaluation value; the random forest algorithm is implemented using the Scikit-learn library in Python, and the model is optimized by adjusting the number of decision trees and the maximum depth, and a comparative analysis is carried out with the weighted superposition result.

8. The ecological sensitivity evaluation method in arid regions according to claim 1, characterized in that: In S5, when refining the sensitivity level, the clustering analysis method is adopted. Based on the data distribution of the comprehensive score, 2-3 sub-levels are added in the highly sensitive and extremely sensitive areas, and the score intervals and ecological characteristics of each sub-level are clarified.

9. The ecological sensitivity evaluation method in arid regions according to claim 1, characterized in that: In S6, the GIS software is used for ecological sensitivity analysis, combined with spatial autocorrelation analysis and buffer analysis, to generate a thematic map of ecological sensitivity; big data mining uses the Hadoop platform to process social media and mobile device location data through the MapReduce programming model.

10. The ecological sensitivity evaluation method in arid regions according to claim 1, characterized in that: In S6, the adaptive management of ecological restoration formulates restoration plans for 1-2 years, 3-5 years, and more than 5 years according to the results of ecological sensitivity evaluation. The ecological indicators in the sensitive areas are monitored every six months, and the restoration strategy is dynamically adjusted according to the monitoring results.

Images