Dynamic monitoring system and method for desertification based on multi-tree species combination prevention and control

Abstract

This invention relates to the field of dynamic monitoring technology for desertification, specifically a dynamic monitoring system and method for desertification based on multi-species combination control. The system includes: a plot layout module, a regional type module, a regional regression module, a state classification module, and a control assessment module. Based on the area of ​​the region to be controlled, multiple plots are divided, and control parameters for each plot are acquired. The control parameters are mapped to the regional types of the plots into multiple state intervals, identifying the transformation trend of each regional type relative to desertification improvement. Based on the transformation trend of the regional types and the rate of change of the current transformation trend, the regional target characteristics of each regional type during desertification improvement are determined. State classification is performed based on the regional target characteristics, and desertification control behaviors are set according to the data proportion under each state classification. Based on the divided desertification control behaviors, the control parameters of the diverse plots are evaluated to obtain a control configuration strategy. This achieves regional adaptability and accuracy in desertification monitoring.

Description

Technical Field

[0001] This invention relates to the field of dynamic monitoring technology for desertification, specifically a dynamic monitoring system and method for desertification control based on the combination of multiple tree species. Background Technology

[0002] Desertification control has long relied on single tree species or mechanical sand fixation measures. However, these measures are susceptible to climate influences, have low coverage efficiency, and suffer from monitoring lag, which reduces the accuracy and practicality of desertification monitoring and affects the precision of remote sensing monitoring of desertification dynamics.

[0003] For example, Chinese Patent Publication No. CN118534086A discloses a dynamic monitoring system for desertification in the Horqin Sandy Land. This system comprises: remote sensing monitoring → ground monitoring → soil sediment content ratio monitoring → desertification comprehensive index → ​​data analysis → early warning system. This invention utilizes Landsat remote sensing data to calculate the vegetation index of the Horqin Sandy Land based on spectral characteristics; it uses a wind erosion model (RWEQ) to calculate the wind erosion modulus of the Horqin Sandy Land; it uses an ASD spectrometer and GF-5 remote sensing data to obtain soil sediment content ratio data, and then uses soil sediment content and land use data to establish a quantitative relationship to calculate the soil sediment content ratio data; it uses the analytic hierarchy process (AHP) to determine the weights of different factors, constructing a comprehensive desertification index that integrates vegetation, dynamics, and soil factors, thereby accurately and objectively monitoring the long-term desertification dynamics of the Horqin Sandy Land.

[0004] For example, Chinese Patent Publication No. CN118258766A discloses a remote sensing monitoring method, system, and device for land desertification in arid areas. The remote sensing monitoring method for land desertification in arid areas includes the following steps: acquiring remote sensing data and soil nutrient data for each monitoring time point within the entire monitoring period; constructing a remote sensing monitoring index system for land desertification in arid areas based on the remote sensing data, obtaining the normalized vegetation index, surface albedo, surface temperature, and temperature-vegetation drought index for each monitoring time point; determining the weight of each monitoring index pixel by pixel based on the spatial heterogeneity of the second law of geography and soil nutrient data; calculating the land desertification index for each monitoring time point based on the monitoring indicators and weights; analyzing the spatiotemporal variation trend of land desertification based on the land desertification index using linear regression analysis; and classifying land desertification levels based on the spatiotemporal variation trend of land desertification.

[0005] Existing technologies describe the use of hierarchical analysis with multiple parameters to explain the comprehensive desertification index with multi-factor weights for early warning of desertification; and the use of surface reflection for spatiotemporal change analysis to determine the level of land desertification. Existing technologies tend to analyze desertification control scenarios based on single parameters such as soil sediment content and normalized vegetation index. However, these data analyses cannot comprehensively analyze the spatial continuity of each data source and the changing trends in the time dimension, resulting in insufficient accuracy in monitoring desertification, difficulty in quantifying the improvement in each region, and limited adaptability of desertification monitoring. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a dynamic monitoring system for desertification prevention and control based on multiple tree species, including: a sample plot layout module, used to divide multiple sample plots with a fixed grid size according to the area of ​​the current area to be controlled, and to acquire the control parameters of the sample plots.

[0007] The region type module is used to map prevention and control parameters to the region type of the sample plots. The mapped data is divided into state intervals for each prevention and control parameter in the form of a long-term series. Based on the transition probability of each prevention and control parameter in the corresponding state interval, the transformation trend of each region type relative to desertification improvement is identified.

[0008] The regional regression module is used to determine the regional target characteristics of a region type during desertification improvement based on the transformation trend of the region type and the rate of change of the current transformation trend.

[0009] The status classification module is used to classify the status based on the regional target characteristics. It divides each regional target characteristic into multiple status categories according to the data types it contains, and sets desertification prevention and control behaviors according to the data proportion under each status category.

[0010] The prevention and control assessment module is used to evaluate prevention and control parameters for diverse areas based on the classified desertification prevention and control behaviors, and to obtain prevention and control configuration strategies.

[0011] A dynamic monitoring method for desertification control based on multi-tree species combination includes: S1, dividing the current area to be controlled into multiple sample plots with a fixed grid size, and acquiring the control parameters of the sample plots.

[0012] S2 maps the prevention and control parameters to the regional types of the sample plots. The mapped data is then divided into state intervals for each prevention and control parameter in the form of a long-term series. Based on the transition probability of each prevention and control parameter in the corresponding state interval, the transformation trend of each regional type relative to desertification improvement is identified.

[0013] S3. Based on the transformation trend of the regional type and the rate of change of the current transformation trend, determine the regional target characteristics of the regional type when desertification is improved.

[0014] S4. Classify the status according to the regional target characteristics. Divide each regional target characteristic into multiple status categories according to the data types it contains. Set desertification prevention and control behaviors according to the data proportion under each status category.

[0015] S5. Based on the classified desertification control behaviors, the control parameters of diverse sites are evaluated to obtain control configuration strategies.

[0016] The beneficial effects of this invention are as follows: First, this invention divides sample plots into fixed grids, integrates four types of prevention and control parameters: normalized vegetation index, surface albedo, soil moisture, and effective soil thickness, and achieves continuous data splicing by binding spatial location with multi-tree species configuration, thereby improving the comprehensiveness of current data coverage and providing a foundation for subsequent analysis of various sample plots.

[0017] Second, this invention establishes a mapping relationship between prevention and control parameters and regional types through technical effects, and divides state intervals based on desertification threshold and improvement threshold. After setting the transition probability and threshold balance for the data in multiple state intervals, it explains the relative situation of current desertification improvement and the quantitative effect of current desertification prevention and control, so as to monitor the dynamic evolution of desertification and identify the improvement situation in desertification prevention and control in advance.

[0018] Third, this invention distinguishes between continuous and discrete boundary regions by determining boundary connectivity, and further describes the transformation trend of desertification improvement by displaying the direction of the boundary regions. It uses the maximum directional boundary length and parameter combination to describe the values ​​of prevention and control parameters after the current desertification improvement, explaining the problem of unclear prevention and control effects caused by ignoring boundary dynamics during desertification treatment. It links the configuration strategy with the location of desertification improvement to describe the data basis for spatial optimization under multi-tree species combination.

[0019] Fourth, this invention categorizes regional target features into dominant, normal, and marginal modes based on similarity metrics. The output content of each mode further illustrates whether the multiple tree species currently configured in the sample plots are suitable for desertification control in the corresponding region. The multiple output modes serve as the basis for scheme configuration allocation during subsequent desertification control optimization, thereby improving the efficiency of desertification control strategy utilization. Finally, by updating the difference before and after, pre-stored instructions are retrieved to generate configuration sub-strategies, and the minimum common subset of the coverage set is calculated. Under the premise of ensuring strategy compatibility, the current configuration strategy can adapt to the terrain differences in various regions, thereby improving the adaptability of the scheme during implementation. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Figure 1This is a schematic diagram of a dynamic monitoring system for desertification control based on the combination of multiple tree species.

[0022] Figure 2 This is a flowchart illustrating the regional type module of a dynamic monitoring system for desertification control based on multi-tree species combinations.

[0023] Figure 3 This is a flowchart illustrating the regional regression module of a dynamic monitoring system for desertification control based on multi-tree species combinations.

[0024] Figure 4 This is a flowchart illustrating the status classification module of a dynamic monitoring system for desertification control based on multi-tree species combinations.

[0025] Figure 5 This is a flowchart illustrating a dynamic monitoring method for desertification control based on the combination of multiple tree species. Detailed Implementation

[0026] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0027] See Figure 1 The dynamic monitoring system for desertification control based on multi-tree species combination includes: a sample plot layout module, a regional type module, a regional regression module, a status classification module, and a control assessment module. The output of the sample plot layout module is connected to the regional type module, the output of the regional type module is connected to the regional regression module, the output of the regional regression module is connected to the status classification module, and the output of the status classification module is connected to the control assessment module.

[0028] The sample plot layout module is used to divide the current area to be treated into multiple sample plots with a fixed grid size, and to obtain the treatment parameters of the sample plots.

[0029] The region type module is used to map prevention and control parameters to the region type of the sample plots. The mapped data is divided into state intervals for each prevention and control parameter in the form of a long-term series. Based on the transition probability of each prevention and control parameter in the corresponding state interval, the transformation trend of each region type relative to desertification improvement is identified.

[0030] The regional regression module is used to determine the regional target characteristics of a region type during desertification improvement based on the transformation trend of the region type and the rate of change of the current transformation trend.

[0031] The status classification module is used to classify the status based on the regional target characteristics. It divides each regional target characteristic into multiple status categories according to the data types it contains, and sets desertification prevention and control behaviors according to the data proportion under each status category.

[0032] The prevention and control assessment module is used to evaluate prevention and control parameters for diverse areas based on the classified desertification prevention and control behaviors, and to obtain prevention and control configuration strategies.

[0033] The control parameters include the Normalized Difference Vegetation Index (NDVI), Albedo, Soil Moisture (Wet), and Effective Soil Depth. These parameters are used to explain the vegetation index of the current vegetation cover area after control, the Albedo index which reflects the surface structure, and the relevant parameters of soil moisture and thickness. They are used to explain the soil preservation effect of the current area under the control of multiple tree species, and whether there is a serious problem of soil erosion, thereby controlling soil desertification.

[0034] Preferably, when dividing the land into diverse sizes, the current area to be treated is displayed as a grid with each 50m section as a unit. Satellite remote sensing images of the area to be treated are acquired, and the Normalized Difference Vegetation Index (NDVI) and Albedo are extracted from the satellite remote sensing images. Soil moisture and effective soil thickness are obtained through soil sampling. Soil moisture is measured by a moisture sensor. Effective soil thickness is determined by the thickness of soil suitable for plant root development. For example, by excavating a soil profile, the distance from the soil surface to the layers unsuitable for plant growth, such as bedrock, hard clay layers, or groundwater level, is observed and measured. This distance is considered as the effective soil thickness. Alternatively, the proportion of sand, silt, and clay particles in soil unsuitable for plant growth can be analyzed. These data are obtained directly from the plant-related data set in the database, based on the type of plant currently being treated.

[0035] Preferably, the implementation of the sample plot layout module also includes: using the spatial location of each sample plot as the basic condition, the tree species combination bound to each sample plot as the configuration condition, and splicing the sample plots according to the time points when the data in the prevention and control parameters change continuously.

[0036] Based on the number of sample plots that are in the same data dimension after being stitched together, set the area type corresponding to each prevention and control parameter.

[0037] The number of sample plots under the same data dimension, as described above, is used to illustrate data at the same point in time and spatial location when data such as Normalized Difference Vegetation Index (NDVI), Albedo, Soil Moisture, and Effective Soil Thickness are continuously changing. The continuously changing time point represents the magnitude of any value among NDVI, Albedo, Soil Moisture, and Effective Soil Thickness changing over time, serving as the subject of subsequent analysis. Using tree species configuration as a condition facilitates the segmentation of sample plots with the same tree species configuration. In this case, the number of tree species may differ in different locations when implementing desertification control, requiring the analysis of sample plots with the same configuration to illustrate the effects on different regional types.

[0038] In one embodiment of the present invention, the region type refers to whether the current location is a type of sand dune, such as a fixed dune, a semi-mobile dune, or a mobile dune, indicating whether the region type at the corresponding location has changed under long-term series control measures. This region type will represent the label marked on the sample plot at the initial stage of control, and this label is used to indicate the progress of improvement and sandification of the current sample plot in multiple monitoring cycles after control.

[0039] like Figure 2 As shown, the implementation of the region type module includes: based on the values ​​of the parameters in the prevention and control parameters, mapping the value range of each parameter in the prevention and control parameters to the region type, and setting multiple region status indices.

[0040] Set the state range corresponding to each region type based on the desertification threshold and improvement threshold pointed to by the region state index.

[0041] It should be noted that the prevention and control parameters will be set in multiple state ranges according to the values ​​of normalized vegetation index, surface albedo, soil moisture and effective soil thickness.

[0042] For example, the Normalized Difference Vegetation Index (NDVI) sets desertification and improvement thresholds based on the current value to illustrate the improvement after the NDVI is clearly identified as desertified and after control measures are implemented. Relevant labels are then set for sample plots at different locations according to the NDVI values. For instance, in the desertification threshold, severe desertification is defined as NDVI ≤ 0.1, corresponding to a vegetation cover of < 30%. This indicates the relative vegetation cover in the area to be controlled. When the vegetation cover is less than 30%, it indicates a significant decrease in vegetation cover in the current area, making it prone to severe desertification. In the desertification threshold, moderate desertification is defined as NDVI between 0.1 and 0.2, simultaneously representing a vegetation cover of 30% to 50%, i.e., using slightly degraded grassland to correspond to moderate desertification. The improvement threshold is then divided into preliminary improvement and significant improvement. Preliminary improvement is defined as NDVI ≥ 0.2, and significant improvement as NDVI ≥ 0.4. In other words, significant improvement is defined as a vegetation cover rate reaching 50%, at which point the control effect is obvious.

[0043] The desertification threshold and improvement threshold of surface albedo are similar to those used in the Normalized Difference Vegetation Index (NDVI), and are also categorized into moderate and severe desertification. For example, severe desertification is defined as Albedo ≥ 0.25, indicating that the surface is bare or sparsely covered, mainly composed of sand or rock, with high albedo. In grasslands and forests, Albedo is typically 0.1–0.2, used to describe bare areas with sparse vegetation cover. Moderate desertification is defined as Albedo between 0.2 and 0.25, indicating the presence of some vegetation cover, but still in a desertified state. The improvement threshold is divided into preliminary improvement (Albedo ≤ 0.2) and significant improvement (Albedo ≤ 0.15). Preliminary improvement indicates increased vegetation cover after desertification control, leading to a decrease in surface albedo. Significant improvement indicates good vegetation recovery and a relatively stable state.

[0044] The NDVI and Albedo values ​​mentioned above can be directly obtained based on existing satellite remote sensing image processing methods, so they will not be elaborated on here. The multiple thresholds related to NDVI and Albedo set above can be set based on historical data from the current area to be treated, classifying it into severely and moderately desertified areas. The data described above is for illustrative purposes only.

[0045] The soil moisture threshold for desertification can be set based on the soil moisture content of desertified land. For example, soil moisture ≤10% can be considered the threshold for severe desertification, and soil moisture between 10% and 15% can be considered the threshold for moderate desertification. Severe desertification indicates insufficient soil moisture due to drought, while moderate desertification represents poor soil water retention and difficulty in maintaining vegetation. The desertification threshold for these conditions can be determined by sampling the Gobi Desert and shifting sand dunes in the current area, and the thresholds for moderate and severe desertification can be adjusted according to the location of the area to be treated. As for the improvement threshold for soil moisture, it can be divided into preliminary improvement (soil moisture ≥15%) and significant improvement (soil moisture ≥20%). Preliminary improvement represents the initial stage of vegetation recovery and an increase in soil moisture. This indicates that the corresponding data in the control parameters have rebounded and improved. Other data can be synchronized to determine whether there is improvement under this regional type. Significant improvement indicates that the soil water retention capacity is enhanced and the plant roots can consolidate water, indicating that multi-species control has a certain effect.

[0046] The desertification threshold and improvement threshold set for effective soil thickness need to be determined according to the soil layer thickness that plants can develop in. For example, soil samples with severe and moderate desertification are taken, and their value ranges are converted into confidence intervals. The upper limit of these intervals is used for differentiation, and a 95% confidence level can be used for the confidence intervals. The desertification threshold for severe desertification can be ≤5cm, and the desertification threshold for moderate desertification can be 5-10cm. These two desertification thresholds represent land with predominantly sandy soil and weak wind resistance. The desertification thresholds are then used to examine whether the current control measures are beneficial to plant growth, based on the initial data collected and subsequent multi-time-point data. As for the improvement thresholds, preliminary improvement with an effective soil thickness ≥10cm and significant improvement with an effective soil thickness ≥15cm can be selected to illustrate the effective soil thickness and stability under the current control measures.

[0047] At this point, based on the different prevention and control parameters collected from each sample plot, and according to the set improvement threshold and desertification threshold, the sample plot will be divided into multiple parameter combination state intervals to illustrate the state of each sample plot relative to the normalized vegetation index, surface albedo, soil moisture and effective soil thickness, and to indicate whether each area is developing normally.

[0048] Based on the range of values ​​of prevention and control parameters within the current state interval under the preset detection cycle, detect the number of prevention and control parameters in each state interval at multiple time points, and set the threshold balance degree of improvement and desertification for the corresponding state interval.

[0049] At this point, the threshold balance can be set based on the balance point method of ROC curves and the dynamic threshold method based on elasticity theory. For example, the balance point method based on ROC curves identifies the maximum difference between the proportion of plots correctly identified as improved and the proportion of plots misjudged as improved in the current area to be identified, thus illustrating the form of dynamic improvement of desertification.

[0050] The dynamic threshold rule based on elasticity theory uses the number of plots in an improved state divided by the total number of plots to indicate the value of the improved state in the current area to be identified. This indicates the threshold balance of the state interval. This threshold balance will dynamically indicate the classification criteria for desertification and improvement, and is used to further monitor the improvement of desertification in the current area. In other words, the current threshold balance represents the proportion of the corresponding parameter in an improved state. When analyzing a single parameter, the threshold balance represents the proportion of that single parameter in an improved state. If it is used to analyze multiple parameters, it represents the proportion of all parameters that are below the improvement threshold.

[0051] The transition probability of each prevention and control parameter is weighted and calculated based on the threshold balance of the state interval, which serves as the indicator value of the transformation trend. The slope corresponding to the indicator value at multiple time points is regarded as the output transformation trend.

[0052] At this point, based on the changes in the corresponding values ​​of each sample plot within a fixed detection period, such as NDVI changing from 0.1 to 0.2, the transition probability of the transformed values ​​is calculated using a Markov chain method. If all four parameters—Normalized Difference Vegetation Index (NDVI), Albedo, Soil Moisture, and Effective Soil Thickness—change at this time, the transition probabilities of these four values ​​are sequentially weighted and summed. The sum of the weighted sum and the threshold balance is considered as the indicator value of the transformation trend at the current time point. The transformation trend is then represented by the slope of the indicator values ​​at multiple time points, indicating an overall tendency for improvement. The weights of NDVI, Albedo, Soil Moisture, and Effective Soil Thickness can be set according to the frequency of each data collection under the identified area's desertification improvement, or by setting fixed values ​​such as 0.3, 0.2, 0.3, and 0.2 sequentially to illustrate the relatively complete trend change pattern of the current area type. The weight of the threshold balance is the ratio of the frequency of the current value to the total frequency in the historical data.

[0053] At this point, by analyzing the trend indicator, we can determine whether the current regional status is dominated by desertification or improvement. A positive slope value indicates continuous improvement, while a negative slope value indicates partial degradation and instability, requiring strengthened environmental prevention and control interventions.

[0054] It should be noted that the above processing will be performed on sample plots under the same regional type, or the changes in regional type will not be considered. The improvement form of the current region under the control will be determined directly based on the areas included in the current area to be controlled. At the same time, the calculation of state transition probability can be displayed on a single sample plot. The slope value corresponding to the transformation trend is presented in the form of a matrix. The changes in the slope value of each sample plot are combined into a matrix, and the value of the matrix in each time period is observed to evaluate its transformation mode. Alternatively, the transformation trend index value can be obtained by weighted summation. After weighted averaging of the normalized vegetation index, surface albedo, soil moisture and effective soil thickness in all sample plots, the four average values ​​are weighted with the threshold balance to obtain the corresponding index value to illustrate the relative desertification improvement.

[0055] Preferably, to make the current transformation trend output more comprehensive, the trends of changes in the four parameters corresponding to the Normalized Difference Vegetation Index (NDVI), Albedo, Soil Moisture, and Effective Soil Thickness will be segmented. For any of these four parameters that shows a change in value, a weighted sum will be applied, and the corresponding parameter will be weighted and summed with the threshold balance of the state interval. At this time, the weights can be set based on the scenario where all four parameters change, so as to obtain the relative parameters at the current desertification improvement. This means that the output transformation trend may represent the improvement of a single parameter or the improvement of multiple parameters combined, so as to more comprehensively represent which parameters have obvious improvement trends after desertification control treatment in the current area, and whether the overall improvement is normal. At this time, the time point for data analysis and collection will be preset to the detection period in months and years to describe the transformation trend at multiple time points.

[0056] Preferably, the implementation method for identifying the conversion trend of a region type further includes: based on the region type corresponding to the conversion trend, when the slope value corresponding to the current conversion trend is positive, outputting the data in the continuous time series corresponding to the current conversion trend.

[0057] When the slope value corresponding to the current transformation trend is negative, the data of the extreme point corresponding to the current transformation trend will be output.

[0058] At this point, the slope value of the curve at each recorded time point will be judged according to the curve combination within each time period, and multiple values ​​related to the current transformation trend will be output. This will enable the most obvious boundary features with the current area type to be found based on the obtained slope value and the actual improvement when judging the improvement of desertification under the transformation trend. That is, to identify the actual effect of multi-tree species prevention and control in areas with desertification such as fixed dunes, semi-mobile dunes and mobile dunes, and then monitor and record the current prevention and control work.

[0059] Preferably, since the current transformation trend will form a composite state interval by crossing the values ​​of multiple prevention and control parameters, and then each sample plot is labeled according to the value range corresponding to the desertification threshold and the improvement threshold, the quantity contained in the state interval is recorded as the proportion of improvement considered to exist under the multiple state intervals divided by the desertification threshold and the improvement threshold, to illustrate the proportion of the current multiple state intervals that are reasonably allocated to the improvement status; the state interval is only used to describe the prevention and control parameters under different value ranges and should be regarded as any description of desertification and improvement.

[0060] In one embodiment of the present invention, when the transformation trend of any parameter in the prevention and control parameters is received, the transformation trend of other parameters is analyzed at the time point of the current transformation trend input. The analysis includes, but is not limited to, identifying the critical points of the change rate of different parameters, as well as each parameter near the critical point. These points are regarded as the target task of the current analysis to identify the value and position when the corresponding trend changes. Using the form of multi-attribute regression and parameter analysis, the regional target characteristics affecting desertification improvement under the current regional type are obtained.

[0061] Under current desertification control measures, a continuous area several kilometers long may be divided into multiple sample plots. Each sample plot is then labeled with its area type. In addition to indicating whether it belongs to mobile dunes, the area type can also be divided into areas with different tree species configurations. Each area is labeled with its area type to illustrate the different desertification improvement effects of multiple tree species combinations. At this point, the length, shape, and spatial continuity of the areas clustered by critical points and mutation points can be used to illustrate the situation under different tree species control measures.

[0062] like Figure 3 As shown, the implementation of the regional regression module includes: for the prevention and control parameters included in the current transformation trend, starting from the rate of change of any one of the prevention and control parameters in the transformation trend, identifying the critical point of each prevention and control parameter in the transformation trend. At this time, the critical point represents the part of the transformation trend where the rate of change exceeds the set threshold. This set threshold can be set to the part that is higher than the moving average and standard deviation of the current overall transformation trend. At this time, the set threshold regards the data that are greater than the average rate of change and twice the standard deviation as its boundary point. The transformation trend formed by the prevention and control parameters of each state interval is analyzed in a centralized manner to determine the change of the transformation trend.

[0063] The regions mapped by each critical point are considered as boundary regions, and the boundary length, shape, and connectivity of each boundary region are obtained. At this point, the regions mapped by each critical point represent the sample plots from which the corresponding control parameters were collected, and these sample plots containing critical points form the boundary regions. The obtained boundary shapes are used to help identify whether each boundary region is connected, and the boundary length and shape are combined to indicate the relative trend under current desertification control, illustrating the effect of using multiple tree species for control.

[0064] When boundary connectivity exists in the boundary region, the current boundary length and boundary shape are traversed, and the direction vector is set by the difference of the transformation trend index values ​​of two adjacent boundary points in the boundary region. Multiple adjacent boundary points are fitted with directions, and the boundary length under the maximum fitted direction is used as the output regional target feature.

[0065] When there is no boundary connectivity in the boundary region, a parameter response plane is established. At this time, a continuous parameter field is set with the values ​​of the prevention and control parameters corresponding to each critical point. Then, the optimal part of the multi-parameter combination is used as the output regional target feature, that is, the optimal combination of each prevention and control parameter under the parameter response plane is used as the output regional target feature.

[0066] Preferably, the implementation method for obtaining boundary connectivity further includes: setting a judgment threshold based on the ratio of the connected area of ​​the mapped boundary region to the total area of ​​the boundary region; when the corresponding ratio is greater than the judgment threshold, the current boundary region is regarded as having boundary connectivity; otherwise, it is regarded as not having boundary connectivity.

[0067] Preferably, the above-mentioned determination of boundary connectivity is based on the ratio of the area of ​​the connected boundary region to the total area of ​​the boundary region. When most boundary regions are connected, the current boundary region is considered to have boundary connectivity. For example, 0.7 is used as the determination threshold at this time. When the area of ​​the connected boundary region accounts for 0.7, it means that the area of ​​desertification improvement and significant change in the currently divided region occupies a relatively large part. Most regions have significant improvement within a specific period. At this time, it is only necessary to look at the length of the boundary region projection in the main improvement direction to know the size of the area corresponding to the current significant improvement, so as to illustrate the trend of desertification improvement and prevention.

[0068] When the value is less than 0.7, it indicates that the areas where significant improvement has occurred are relatively scattered and located in multiple locations. In this case, it is necessary to obtain the parameter combinations distributed in multiple locations, and to perform regression analysis or other forms of analysis on these parameter combinations to find the optimal parameter combination with the smallest error over multiple iterations. This optimal combination is then used as the output to illustrate the values ​​of prevention and control parameters in areas where significant improvement has occurred, in order to capture local heterogeneity and adapt to the treatment methods for desertification prevention and control of multiple tree species under complex conditions.

[0069] Similarly, the judgment threshold can be set based on the confidence interval of the ratio of the connected area of ​​the boundary region to the total area of ​​the boundary region under the current boundary connectivity analysis. That is, the range of data in the historical data under the corresponding area ratio is set with a 95% confidence level confidence interval, and the median value of the confidence interval is used as the current judgment threshold.

[0070] Preferably, the above-mentioned direction vector is used as the length of its direction vector by taking the difference of the index values ​​of the transformation trend of the two boundary points, and the angle of the line segment connecting the two boundary points as its angle. Then, multiple direction vectors are fitted in the form of principal component analysis, the covariance matrix and the corresponding eigenvalues ​​are calculated, and the direction with the largest variance is regarded as the current maximum direction of the fit. Then, the length value of the boundary region projected onto this direction is regarded as the output regional target feature.

[0071] Preferably, the optimal combination of each prevention and control parameter is achieved by taking soil moisture as the dependent variable and other prevention and control parameters as independent variables, and taking the set of prevention and control parameters with the smallest error under linear regression calculation as the current optimal combination.

[0072] The main reason for choosing soil moisture at this time is that soil moisture is a key indicator of vegetation survival, which directly affects NDVI and the corresponding surface albedo. As for soil thickness, although it represents the relative conditions for root growth, it can only play a relative role when the soil moisture is high. That is, the predicted value is calculated by linear regression using data from multiple boundary areas. The error between the predicted value and the actual soil moisture value in the corresponding boundary area is calculated, and the area with the smallest error is found. This area will represent the area where desertification control is most obvious under the current multi-tree species combination.

[0073] In one embodiment of the present invention, when classifying the state, the proportion of various data to the total data after using multiple tree species for treatment under different desertification forms is described, the influence of the main tree species on desertification control after growth is explained, and the desertification boundary after the influence is distinguished, and the desertification effect under multiple tree species control is analyzed.

[0074] like Figure 4 As shown, the implementation of the state classification module includes: using regional target features as candidate classification points, using the similarity measure between candidate classification points and prevention parameters for state classification, setting the state classification corresponding to the candidate classification points, and using the classification points corresponding to the regional target features to divide the collected data into multiple modes to describe multiple sets of descriptions under the current regional target features; since the data output by the regional target features has two forms: the projection boundary length and the optimal parameter combination, the optimal parameter combination can be used to divide the parameters under the optimal soil moisture change in multiple time situations as candidate classification points, and the boundary length will be the boundary point fitted when obtaining the boundary length, and its corresponding boundary point will be used as candidate classification points to divide the value situation under various desertification improvement conditions. These data and the collected prevention parameters are classified according to the similarity measure. Each time classification is performed, the corresponding data and the currently collected data are clustered using the prevention parameters corresponding to the current candidate classification point to adapt to the dynamic changes of multiple parameters under desertification prevention.

[0075] At this point, cosine similarity is used to calculate the similarity between the value of the current candidate classification point and other collected prevention and control parameters to obtain a similarity measure. During clustering, the data with a cosine similarity greater than 0.6 are combined with multiple groups of data that are obviously similar to the current candidate classification point to discover the dynamic behavior pattern of governance under multiple tree species combinations in the data.

[0076] The data proportion under each state category is derived, and the data proportion is compared with the preset mode range. The current state is classified into dominant mode, normal mode and marginal mode in turn.

[0077] The data with the largest proportion in the state classification is regarded as the dominant mode, and the candidate classification points of the dominant mode are used as the output desertification control behavior. The dominant mode mainly outputs the core parameter combination, which describes the main improvement forms under the current desertification control, and this data serves as the basis for subsequent adjustments to the intervention methods.

[0078] Data falling within the preset mode range is considered the normal mode, and the direction of each prevention and control parameter under the normal mode across multiple time periods is considered the output desertification prevention and control behavior. The direction across multiple time periods mentioned here means that the corresponding parameter is in a state of increase, decrease, etc., over a continuous time period. This can be achieved through time series analysis, using linear regression to indicate the slope value and trend term of the current prevention and control parameter. These contents are used as the output desertification prevention and control behavior to illustrate the periodic changes under the current state classification.

[0079] When the value is less than the lower limit of the preset mode range, the data in the corresponding state category is regarded as an edge mode. The state transition of each control parameter in the edge mode is used as the output of the desertification control behavior in a chain structure. At this time, the state transition of the control parameter is recorded to illustrate the change of each control parameter under desertification control. It tends to use a chain structure to illustrate that NDVI is less than 10% → NDVI recovers to the baseline value, or NDVI is less than 10% → NDVI rebounds. The chain structure records the intervention content experienced when the control parameter changes. This part is used as the content of rapid response to degradation risk. It focuses on identifying whether the corresponding control parameter under the chain structure has reached the return to normal state. The output of the chain structure will include the actual behavior implemented under the current multi-tree species control, such as the recovery status of control parameters under the measures of drone watering, seeding robot replanting, etc.

[0080] It should be noted that when using candidate classification points for classification, a corresponding text label will be set for each candidate classification point, and candidate classification points under the same description will be regarded as the same state classification.

[0081] Preferably, the currently set preset mode range will be based on the current input state classification, and multiple value ranges will be set sequentially. For example, 15%-40% can be selected as the current preset mode range, or the conventional form of multi-tree species combination under desertification control can be selected from historical data. The amount of data contained in the conventional form each time can be expressed using confidence intervals, and the upper and lower limits of the confidence interval at a 95% confidence level can be used as the preset mode range at this time to represent the relative behavior of desertification control mainly within the corresponding range.

[0082] The dominant mode then explains that the patterns are highly similar in most cases, which is the main reason for choosing strategies under desertification control. The normal mode tends to acquire specific environmental data and uses multi-time change trajectories for explanation. Here, the multi-time indicator is used to represent the time change trajectory to explain the changes under the normal mode. The edge mode will represent the highlighted mutation content, which is presented in the form of a state chain to facilitate subsequent processing of the mutation and prevent the desertification from worsening.

[0083] In one embodiment of the present invention, when setting the prevention and control configuration strategy, it is necessary to describe the behavior represented by the output desertification prevention and control behavior using different mode representation forms, such as rapid improvement type, gradual stabilization type, fluctuation maintenance type and degradation risk type, to describe the relative behavior that can be generated under the current multi-tree species combination, thereby transforming it into the corresponding strategy.

[0084] The rapid improvement type is represented by short-cycle parameter jumps, the gradual stabilization type represents long-cycle linear changes, the fluctuation maintenance type represents threshold range oscillations, and the degradation risk type represents the decay of key parameters. In this case, the dominant mode and the marginal mode included in the state classification will be used as the main monitoring content for the fluctuation maintenance type and the degradation risk type. The configuration strategies that need to be set are viewed according to these two types. As for the normal mode, the effectiveness of prevention and control is evaluated through the rapid improvement type and the gradual stabilization type, and the corresponding configuration strategies are set. These configuration strategies will be set in the database and directly compared and extracted based on the parameters of the currently input desertification prevention and control behavior.

[0085] The implementation of the prevention and control assessment module includes: when the current desertification prevention and control behavior is in an update state, according to the time interval of each update, the difference between the desertification prevention and control behavior before and after the update is viewed. At this time, the difference between the desertification prevention and control behavior represents the difference of the corresponding prevention and control parameters; using the difference between the desertification prevention and control behavior, the pre-stored information instructions are retrieved to form the configuration sub-strategy of the current desertification prevention and control behavior.

[0086] When the current desertification control behavior is not in an updated state, obtain the configuration sub-strategy based on the control parameter range of the current desertification control behavior; calculate the coverage set of each configuration sub-strategy, and use the least common subset in the coverage set as the output control configuration strategy.

[0087] Preferably, when obtaining the configuration strategy for the current desertification control behavior, the strategy is extracted based on the state classification corresponding to the desertification control behavior. Then, according to the state classification, configuration strategies for normalized vegetation index, surface albedo, soil moisture, and effective soil thickness are retrieved from the database. Configuration strategies consistent with the current state classification are retrieved, and subsequent configuration sub-strategies are formed according to the pre-stored information instructions contained in the configuration strategy. The pre-stored information instructions explain how to intervene in the current multi-tree species control efforts, setting relevant instructions based on the specific values ​​of the control parameters.

[0088] Preferably, when calculating the coverage set of each configuration sub-strategy, the set of parameters covered by each configuration sub-strategy is regarded as the coverage set. Then, the strategy corresponding to the least common subset in the coverage set is obtained as the subsequent output prevention and control configuration strategy, so as to achieve the treatment of desertification prevention and control while satisfying the common characteristics of the overall desertification prevention and control behavior.

[0089] Preferably, the above-mentioned update status indicates that the data in the currently acquired desertification prevention and control behavior has been updated, representing the dynamic change in the desertification improvement and governance trend, and a configuration sub-strategy is generated according to each dynamic change to deal with the problem of missing real-time response, so as to realize real-time tracking of desertification improvement.

[0090] like Figure 5 As shown, the present invention also provides a dynamic monitoring method for desertification control based on multi-tree species combination, including: S1, dividing multiple sample plots with a fixed grid size according to the area of ​​the current area to be controlled, and acquiring the control parameters of the sample plots.

[0091] S2 maps the prevention and control parameters to the regional types of the sample plots. The mapped data is then divided into state intervals for each prevention and control parameter in the form of a long-term series. Based on the transition probability of each prevention and control parameter in the corresponding state interval, the transformation trend of each regional type relative to desertification improvement is identified.

[0092] S3. Based on the transformation trend of the regional type and the rate of change of the current transformation trend, determine the regional target characteristics of the regional type when desertification is improved.

[0093] S4. Classify the status according to the regional target characteristics. Divide each regional target characteristic into multiple status categories according to the data types it contains. Set desertification prevention and control behaviors according to the data proportion under each status category.

[0094] S5. Based on the classified desertification control behaviors, the control parameters of diverse sites are evaluated to obtain control configuration strategies.

[0095] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A dynamic monitoring system for desertification control based on multi-tree species combinations, characterized in that, include: The sample plot layout module is used to divide the current area to be treated into multiple sample plots with a fixed grid size, and to obtain the treatment parameters of the sample plots; The region type module is used to map prevention and control parameters to the region type of the sample plot. The mapped data is divided into state intervals for each prevention and control parameter in the form of a long-term series. Based on the transition probability of each prevention and control parameter in the corresponding state interval, the transformation trend of each region type relative to desertification improvement is identified. The regional regression module is used to determine the regional target characteristics of a region type during desertification improvement based on the transformation trend of the region type and the rate of change of the current transformation trend. The status classification module is used to classify the status based on the regional target characteristics. It divides each regional target characteristic into multiple status categories according to the data types it contains, and sets desertification prevention and control behaviors according to the data proportion under each status category. The prevention and control assessment module is used to evaluate prevention and control parameters for diverse areas based on the classified desertification prevention and control behaviors, and to obtain prevention and control configuration strategies.

2. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The control parameters include normalized vegetation index, surface albedo, soil moisture and effective soil thickness, where effective soil thickness is the thickness of the soil suitable for plant root development.

3. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The implementation methods for the sample plot layout module also include: The spatial location of each sample plot is used as the basic condition, and the tree species combination bound to each sample plot is used as the configuration condition. The sample plots are spliced ​​together according to the time points when the data in the prevention and control parameters change continuously. Based on the number of sample plots that are in the same data dimension after being stitched together, set the area type corresponding to each prevention and control parameter.

4. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The implementation methods of the region type module include: Based on the values ​​of the parameters in the prevention and control parameters, the mapping relationship between the value range of each parameter in the prevention and control parameters and the regional type is set, and multiple regional status indices are set. Set the state range corresponding to each region type based on the desertification threshold and improvement threshold pointed to by the region state index; Based on the range of values ​​of prevention and control parameters within the current state interval under the preset detection cycle, detect the number of prevention and control parameters in each state interval at multiple time points, and set the threshold balance degree of improvement and desertification for the corresponding state interval. The transition probability of each prevention and control parameter is weighted and calculated based on the threshold balance of the state interval, which serves as the indicator value of the transformation trend. The slope corresponding to the indicator value at multiple time points is regarded as the output transformation trend.

5. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 4, characterized in that, Other methods for identifying conversion trends across different region types include: Based on the region type corresponding to the conversion trend, when the slope value corresponding to the current conversion trend is positive, the data within the continuous time series corresponding to the current conversion trend will be output. When the slope value corresponding to the current transformation trend is negative, the data of the extreme point corresponding to the current transformation trend will be output.

6. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The implementation methods of the region regression module include: For the prevention and control parameters included in the current conversion trend, starting from the rate of change of any one of the prevention and control parameters in the conversion trend, the critical point of each prevention and control parameter in the conversion trend is identified. The regions mapped by each critical point are considered as boundary regions, and the boundary length, boundary shape, and boundary connectivity of each boundary region are obtained. When boundary connectivity exists in the boundary region, the current boundary length and boundary shape are traversed, and the direction vector is set by the difference of the transformation trend index value of two adjacent boundary points in the boundary region. The direction of multiple adjacent boundary points is fitted, and the boundary length under the maximum fitted direction is used as the output regional target feature. When there is no boundary connectivity in the boundary region, a parameter response plane is established, and the optimal combination of each prevention and control parameter under the parameter response plane is used as the output regional target feature.

7. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 6, characterized in that, Other ways to obtain boundary connectivity include: A judgment threshold is set based on the ratio of the connected area of ​​the mapped boundary region to the total area of ​​the boundary region. When the corresponding ratio is greater than the judgment threshold, the current boundary region is considered to have boundary connectivity; otherwise, it is considered to have no boundary connectivity.

8. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The implementation methods of the state classification module include: Regional target features are used as candidate classification points. The similarity measure between the candidate classification points and prevention and control parameters is used to classify the status. The status classification corresponding to the candidate classification points is set. The data proportion under each state category is derived, and the data proportion is compared with the preset mode range. The current state is classified into dominant mode, normal mode and marginal mode in turn. The data with the largest proportion in the state classification is regarded as the dominant mode, and the candidate classification points of the dominant mode are used as the output desertification prevention and control behavior. Data belonging to the preset mode range is regarded as the normal mode, and the direction of each prevention and control parameter under the normal mode at multiple times is regarded as the output desertification prevention and control behavior. When the value is less than the lower limit of the preset mode range, the data in the corresponding state category is regarded as the edge mode. The state transition of each prevention and control parameter in the edge mode is used as the output of desertification prevention and control behavior in a chain structure.

9. The dynamic monitoring system for desertification control based on multi-tree species combination as described in claim 1, characterized in that, The implementation methods of the prevention and control assessment module include: When the current desertification control behavior is in the update state, according to the time interval of each update, check the difference between the desertification control behavior before and after the update, use the difference in desertification control behavior to retrieve the pre-stored information instructions, and form the configuration sub-strategy of the current desertification control behavior. When the current desertification control behavior is not in an updated state, obtain the configuration sub-strategy based on the control parameter range of the current desertification control behavior; calculate the coverage set of each configuration sub-strategy, and use the least common subset in the coverage set as the output control configuration strategy.

10. A dynamic monitoring method for desertification control based on multi-tree species combinations, characterized in that, include: S1, Based on the area of ​​the current area to be treated, divide it into multiple sample plots with a fixed grid size, and obtain the treatment parameters of the sample plots; S2, map the prevention and control parameters to the regional types of the sample plots, divide the mapped data into state intervals for each prevention and control parameter in the form of a long-term series, and identify the transformation trend of each regional type relative to desertification improvement based on the transition probability of each prevention and control parameter in the corresponding state interval. S3. Based on the transformation trend of the regional type and the rate of change of the current transformation trend, determine the regional target characteristics of the regional type when desertification is improved. S4. Classify the status according to the regional target characteristics. Divide each regional target characteristic into multiple status categories according to the data types it contains. Set desertification prevention and control behaviors according to the data proportion under each status category. S5. Based on the classified desertification control behaviors, the control parameters of diverse sites are evaluated to obtain control configuration strategies.

Images