Space-time analysis method, device and equipment for plateau wetland disturbance mechanism

Abstract

The invention discloses a space-time analysis method, device and equipment for a plateau wetland disturbance mechanism, relates to the technical field of wetland data analysis, and is used for solving the problem of poor precision of a gray prediction model caused by external disturbance factors in the prior art. Comprising the following steps: acquiring long-time-sequence basic data, and calculating an emerging hotspot value of each statistical region by adopting a preset clustering method; determining a trend result of time sequence prediction based on the emerging hotspot value; calculating an abnormal value in the long-time-sequence wetland map, and carrying out spatial clustering and abnormal detection on a salient region in a neighborhood to obtain an abnormal change condition; based on a geographical weighted regression model, adding a time label, and analyzing the correlation between the wetland change area of different long-time-sequence regions and the night light change; building a self-adaptive multi-dimensional gray scale prediction model, and determining disturbance factors influencing the area change of the wetland; and performing space-time analysis on the wetland change based on the trend result, the abnormal change condition, the correlation result and the disturbance factor.

Description

Technical Field

[0001] The present invention relates to the technical field of wetland data analysis, and in particular to a method, device and equipment for spatiotemporal analysis of plateau wetland disturbance mechanisms. Background Art

[0002] Wetlands are one of the most important and valuable ecosystems on Earth. They play a unique role in improving water quality, alleviating floods and droughts, and providing habitats for plants and animals. This ecosystem with unique functions is an important part of the Earth's ecological environment and is known as the "Kidney of the Earth." Recent experience and research have shown that wetlands account for nearly 35% of the carbon reserves of terrestrial ecosystems, are an important part of the global carbon cycle, and are also strategic resources related to national and regional ecological security in the process of achieving sustainable development.

[0003] As part of the wetland ecosystem, plateau wetlands and urban wetlands play irreplaceable roles in maintaining biodiversity, flood storage and drought prevention, pollution degradation, and greenhouse gas deposition due to their unique geographical location and socio-economic role. However, wetland ecosystems are extremely fragile, and the combined effects of natural and human factors have led to increasingly obvious wetland degradation. Globally, with the continuous growth of population and the further development of urbanization, many wetlands are facing the continuous threat of resource exploitation or depletion. Therefore, the study and protection of wetlands has become one of the issues that have attracted widespread attention and research worldwide. However, there are few studies on plateau wetlands and urban wetlands that reveal the impact of human disturbance. Existing wetland spatiotemporal analysis research methods mainly focus on the time series changes of wetlands or the mechanisms of wetland changes at a local scale. These methods can reveal detailed response patterns of wetland change information, but they have limitations in long-term macroscopic observations. It is worth mentioning that the maps and driving mechanisms of the time series changes of large-area plateau wetlands are of great significance for predicting carbon budgets, global change models, estimating the productivity of the entire ecosystem, and estimating economic impacts.

[0004] Exploring the spatiotemporal dynamics of plateau wetlands requires the support of long-term time series data, and GIS technology with the advantages of continuous observation and multiple spatiotemporal scales provides an effective means for this. In addition, the structure of the plateau wetland ecosystem itself is complex and stable. In addition to the interference of climate factors, with the strengthening of human activities such as economic development and accelerated urbanization, over-exploitation and utilization of wetland landscape areas, the landscape area of ​​plateau wetlands has been transformed drastically, and the consequences of wetland function degradation have become increasingly serious. In order to further explore the coupling relationship between plateau wetlands and external interference and natural factors, it is necessary to combine the perspectives of plateau wetlands and urban wetlands to conduct a systematic mechanism analysis at a more macroscopic scale. Space and time are the two basic dimensions that provide a framework for all human activities, social events, and environmental processes. Spatiotemporal modeling has also been an important research hotspot in the field of geographic information science (GIScience). However, enhancing the spatiotemporal analysis and modeling capabilities in the current GIS environment remains a major challenge, especially in the era of big data. New methods need to be developed to encourage spatiotemporal thinking and discover useful spatiotemporal information and knowledge in spatiotemporal data to better understand social and environmental dynamics.

[0005] Therefore, there is an urgent need to provide a more reliable spatiotemporal analysis scheme for the disturbance mechanisms of plateau wetlands. Summary of the invention

[0006] The purpose of the present invention is to provide a spatiotemporal analysis method, device and equipment for plateau wetland disturbance mechanism, which is used to solve the problem of poor accuracy of grey prediction model due to external disturbance factors in the prior art.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a method for spatiotemporal analysis of plateau wetland disturbance mechanisms, the method comprising:

[0009] Acquire long-term basic data; the long-term basic data at least includes long-term wetland maps, long-term nighttime light satellite data, long-term annual precipitation data, and annual temperature and climate data;

[0010] The emerging hotspot value of each statistical area is calculated by using a preset clustering method; and the trend result of the time series prediction is determined based on the emerging hotspot value; the statistical area is determined based on the long-time series wetland map;

[0011] Calculate the outliers in the long-term wetland map, and perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes;

[0012] Based on the geographically weighted regression model, the time label is added to analyze the correlation between the wetland change area and night light changes in different regions of the long time series, and obtain the correlation results;

[0013] Build an adaptive multi-dimensional grayscale prediction model to determine the disturbance factors that affect wetland area changes;

[0014] A spatiotemporal analysis of wetland changes was conducted based on the trend results, abnormal changes, correlation results, and disturbance factors.

[0015] Optionally, before calculating the emerging hotspot value of each statistical area in the plurality of statistical areas by using a preset clustering method, the method further includes:

[0016] Determining a target statistical step length based on the long time series wetland map in the long time series basic data;

[0017] A plurality of statistical areas are determined based on the target statistical step size.

[0018] Optionally, the method of calculating the emerging hotspot value of each statistical area in the plurality of statistical areas by using a preset clustering method; and determining the trend result of the time series prediction based on the emerging hotspot value, comprises:

[0019] Using a hotspot analysis clustering method to calculate the emerging hotspot value of each of the plurality of statistical areas based on a preset range and the target statistical step;

[0020] Compare the emerging hotspot values ​​of the previous cycle with the emerging hotspot values ​​of the second cycle;

[0021] If the value of the emerging hotspot in the previous cycle is less than the value of the emerging hotspot in the second cycle, the result is +1;

[0022] If the value of the emerging hotspot in the previous cycle is greater than the value of the emerging hotspot in the second cycle, the result is -1;

[0023] If the value of the emerging hotspot in the previous cycle is equal to the value of the emerging hotspot in the second cycle, the result remains unchanged, and the final result value is generated after calculating the data of all time periods in each of the statistical areas;

[0024] The emerging hotspot values ​​of each time period are integrated with the MannKendall trend results to obtain the trend results of time series forecasting.

[0025] Optionally, the calculating of outliers in the long-term wetland map, and spatially clustering and detecting outliers for significant areas in the neighborhood based on the outliers to obtain abnormal changes, includes:

[0026] Use Local Moran's I to calculate z-values ​​spatially;

[0027] Based on the z value, the significant areas in the neighborhood are spatially clustered and anomaly detected to generate clustering results;

[0028] According to the clustering results, abnormal changes are tested on the basis of time series.

[0029] Optionally, perform spatial clustering and anomaly detection on the significant areas in the neighborhood based on the z-values ​​to generate clustering results, including:

[0030] For any of the statistical regions, if the z value of the statistical region is a positive value higher than the first preset threshold, it means that the regions around the statistical region are similar values; if the regions around the statistical region are all high values, the clustering result is a high-high cluster; if the regions around the statistical region are all low values, the clustering result is a low-low cluster;

[0031] If the z value of the statistical area is a negative value lower than the second preset threshold, it means that there is a large difference between the statistical area and the surrounding area, and it is a statistically significant spatial data outlier; the second preset threshold is smaller than the first preset threshold.

[0032] Optionally, the geographically weighted regression model is based on adding time tags to analyze the correlation between the wetland change area and night light changes in different regions of the long time series to obtain correlation results, including:

[0033] A geographically weighted regression model is used to model the change in wetland area and the intensity of night light changes in a preset area, and the results are classified according to the correlation coefficient to obtain classification results; the classification results are used to indicate the degree of influence of disturbance factors on wetland area changes in different regional areas.

[0034] Optionally, the adaptive multi-dimensional grayscale prediction model is constructed to determine disturbance factors affecting wetland area changes, including:

[0035] Select the time window;

[0036] In combination with the relevant variables in each time window, the target coefficient of the adaptive multidimensional grayscale prediction model is calculated; the adaptive multidimensional grayscale prediction model is a model trained based on the long-term wetland map, the long-term nighttime light satellite data, the long-term annual precipitation data and the annual temperature and climate data;

[0037] Select the time window and related variables corresponding to the minimum target coefficient;

[0038] Determine the target key climatic factors and disturbance factors that affect the wetland area in the corresponding region.

[0039] Compared with the prior art, the present invention provides a spatiotemporal analysis method for the disturbance mechanism of plateau wetlands. By obtaining long-term basic data, a preset clustering method is used to calculate the emerging hotspot value of each statistical area; and the trend results of time series prediction are determined based on the emerging hotspot values; the outliers in the long-term wetland map are calculated, and the significant areas in the neighborhood are spatially clustered and anomaly detected based on the outliers to obtain abnormal changes; based on the geographically weighted regression model, time labels are added to analyze the correlation between the wetland change area and night light changes in different regions in the long time series to obtain correlation results; an adaptive multidimensional grayscale prediction model is built to determine the disturbance factors that affect the change in wetland area; and the wetland changes are spatiotemporally analyzed based on trend results, abnormal changes, correlation results and disturbance factors. The technical solution of the present invention incorporates the time dimension into the geographically weighted regression analysis method, and optimizes the adaptive multidimensional grayscale prediction model at the same time, so as to reveal the driving mechanism of interference factors on plateau urban wetlands and construct a geographical spatiotemporal mining system; the spatiotemporal changes of wetland data are detected by performing emerging spatiotemporal hotspot analysis and local outlier analysis on long-term plateau wetlands, and the spatiotemporal geographically weighted regression analysis is used to analyze the night light data and wetland classification data to determine the correlation between interference factors and wetland area changes; finally, the adaptive multidimensional grayscale prediction model method is used to improve the model accuracy, so that the optimized AGM (1, n) model has better simulation and prediction performance.

[0040] In a second aspect, the present invention provides a spatiotemporal analysis device for plateau wetland disturbance mechanisms, the device comprising:

[0041] A long-term basic data acquisition module is used to acquire long-term basic data; the long-term basic data at least includes a long-term wetland map, long-term nighttime light satellite data, long-term annual precipitation data, and annual temperature and climate data;

[0042] An emerging hotspot value calculation module is used to calculate the emerging hotspot value of each statistical area using a preset clustering method; and determine the trend result of the time series prediction based on the emerging hotspot value; the statistical area is determined based on the long-time series wetland map;

[0043] An outlier calculation module is used to calculate outliers in the long-term wetland map, and to perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes;

[0044] A correlation analysis module is used to analyze the correlation between the wetland change area and night light change in different regions of the long time series based on a geographically weighted regression model and adding a time tag to obtain a correlation result;

[0045] Adaptive multi-dimensional grayscale prediction model building module, used to build an adaptive multi-dimensional grayscale prediction model to determine the disturbance factors that affect the change of wetland area;

[0046] The spatiotemporal analysis module is used to perform spatiotemporal analysis on wetland changes based on the trend results, abnormal changes, correlation results and disturbance factors.

[0047] In a third aspect, the present invention provides a spatiotemporal analysis device for plateau wetland disturbance mechanisms, the device comprising:

[0048] A memory, a processor and a communication interface coupled to the processor; the memory stores a computer program that can be run by the processor; when the processor runs the computer program, it executes the above-mentioned spatiotemporal analysis method of a plateau wetland disturbance mechanism.

[0049] In a fourth aspect, the present invention provides a computer storage medium having instructions stored therein, which, when executed, implement the above-mentioned spatiotemporal analysis method of a plateau wetland disturbance mechanism.

[0050] The technical effects achieved by the device type solution provided in the second aspect, the equipment type solution provided in the third aspect, and the computer storage medium solution provided in the fourth aspect are the same as those of the method type solution provided in the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0052] Figure 1 A schematic diagram of the process of spatiotemporal analysis of plateau wetland disturbance mechanisms;

[0053] Figure 2 A schematic diagram of the overall implementation framework for spatiotemporal analysis of plateau wetland disturbance mechanisms;

[0054] Figure 3 Schematic diagram of time sliding window and variable selection for AGM(1,n);

[0055] Figure 4 It is a schematic diagram of the distribution of emerging hot spots and space-time cubes in some regions;

[0056] Figure 5 It is a local outlier analysis of wetland area changes in multiple types of areas and a schematic diagram of space-time cubes;

[0057] Figure 6 This is a schematic diagram of the SGWR related grade distribution;

[0058] Figure 7 It is a schematic diagram of the comparative analysis between local outlier analysis and SGWR;

[0059] Figure 8 This is a schematic diagram of the structure of a spatiotemporal analysis device for plateau wetland disturbance mechanisms;

[0060] Fig. 9 Schematic diagram of the structure of a spatiotemporal analysis device for plateau wetland disturbance mechanisms. DETAILED DESCRIPTION

[0061] In order to clearly describe the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, words such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. For example, the first threshold and the second threshold are only used to distinguish different thresholds, and their order is not limited. Those skilled in the art can understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit them to be different.

[0062] It should be noted that, in the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0063] In the present invention, "at least one" means one or more, and "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b, c can be single or multiple.

[0064] Commonly used time series analysis methods include principal component analysis, R / S method, Fourier wavelet analysis, Pearson correlation coefficient analysis, etc., while the Mann-Kendall nonparametric test method, which is more suitable for non-normal distribution data, is relatively less used. In addition, due to the widespread spatial heterogeneity, the traditional least square regression model (ordinary least square regression, OLSR) can no longer meet the current needs of spatial correlation analysis. Of course, in addition to space, time is also an important aspect related to social activities and environmental processes. Many variables of interest in earth science are not only observed on a spatial scale, but also can provide valuable information about the dynamics of underlying spatial processes on a temporal scale, thereby enabling the prediction of related variables. However, due to the imperfect monitoring and management system, the loss and lack of historical data make it difficult to expand the analysis on the temporal scale. In order to improve the above problems, it is necessary to incorporate the time dimension into spatial regression and find a more suitable spatiotemporal prediction model. On the one hand, based on the consideration of spatial heterogeneity, the Geographically Weighted Regression model (Geographically Weighted Regression, GWR) adopts the spatial weight matrix method to introduce the idea of ​​spatially variable coefficients, making the spatial sample points that are traditionally assumed to be independent become interdependent. To a certain extent, it reflects the characteristics of spatial locality. However, the GWR algorithm mainly models the relationship between geospatial data within the same time slice, which is somewhat insufficient for geospatial data with time tags. On the other hand, in order to solve the uncertainty problem caused by human cognitive limitations and internal and external interference factors in the system, the existing technology focuses on different angles and proposes various unique theoretical methods, such as fuzzy mathematics based on experience, rough set theory with the help of information tables, gray system theory for small data, and probability statistics based on large samples. Among them, the gray system theory, which is mainly aimed at the problem background characterized by small samples and poor information, can predict the development trend of data without long time series, and has attracted the attention and recognition of many famous scholars at home and abroad. However, the model does not have a quantitative method for selecting time windows and screening related variables. If the time series window is too long, it is quite different from recent data, and the variables are weakly correlated with a certain area, which will lead to poor accuracy of the gray prediction model.

[0065] In order to solve the problems in the prior art, the present invention proposes a spatiotemporal analysis method, device and equipment for plateau wetland disturbance mechanism, which is a general and innovative geographic spatiotemporal mining system to simulate and analyze the change process and response scheme of long-term plateau urban wetlands, incorporate the time dimension into the geographical weighted regression analysis method, and optimize the adaptive multidimensional grayscale prediction model at the same time, revealing the driving mechanism of external disturbance on plateau urban wetlands while constructing a geographic spatiotemporal mining system. The spatiotemporal changes of wetland data are detected by performing emerging spatiotemporal hot spot analysis (Emerging Hot Spot Analysis) and local outlier analysis (Local Outlier Analysis) on long-term plateau wetlands. Then, the spatiotemporal geographically weighted regression analysis (Spatiotemporal Geographically Weighted Regression, SGWR) is used to analyze the nighttime light data (Nighttime light, NTL) and wetland classification data to explore the correlation between external disturbance and wetland area changes. Finally, the adaptive multidimensional grayscale prediction model (AGM (1, n)) method is used to establish a wetland area prediction model using NTL, annual temperature data and annual precipitation data.

[0066] Next, the solution provided by the embodiments of this specification is described in conjunction with the accompanying drawings:

[0067] like Figure 1 As shown, the process may include the following steps:

[0068] Step 110: Obtain long time series basic data;

[0069] The long-term basic data at least include long-term wetland maps, long-term nighttime light satellite data, long-term annual precipitation data and annual temperature and climate data.

[0070] Step 120: Calculate the emerging hotspot value of each statistical area using a preset clustering method; and determine the trend result of the time series prediction based on the emerging hotspot value; the statistical area is determined based on the long-term wetland map.

[0071] The preset clustering method can be the GetisOrd statistics clustering method. For long-term wetland maps, determine the appropriate statistical step size and use GetisOrd Statistics clustering method, calculate each statistical area Then, the The value is integrated with the MannKendall trend result to obtain the trend result of time series forecasting.

[0072] Emerging hotspot analysis was performed based on long-term wetland maps using GetisOrd Statistics calculation On this basis, Mann-Kendall is used to predict the trend. The pixels with positive value increase are accumulated positively. Pixels with negative value growth are accumulated as negative values ​​to obtain the trend result of time series prediction.

[0073] Step 130: Calculate the outliers in the long-term wetland map, and perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes.

[0074] Local Moran's I can be used to calculate the z value spatially, perform spatial clustering and anomaly detection on the significant areas in the neighborhood, and generate clustering results.

[0075] Outlier analysis is aimed at long-term wetland maps. Local Moran's I is used to calculate the value of each unit cell to determine the clustering type of each unit cell. On this basis, Mann-Kendall is used for trend prediction to accumulate the category accumulation in the long time series.

[0076] Step 140: Based on the geographically weighted regression model, a time label is added to analyze the correlation between the wetland area change in different regions of the long time series and the night light change to obtain the correlation result.

[0077] Geographically Weighted Regression model (GWR).

[0078] This step is to add time tags on the basis of the GWR algorithm, and then conduct a correlation analysis between the wetland area and night light changes in different time series.

[0079] Adding time tags based on the geographically weighted regression model can not only reflect the spatial location characteristics of the model but also take into account the time factor, which can effectively help analyze the spatiotemporal characteristics of the data. The correlation between the change in wetland area and night light changes in different regions over a long time series is analyzed.

[0080] Step 150: Build an adaptive multi-dimensional grayscale prediction model to determine the disturbance factors that affect the change of wetland area.

[0081] Grey Model (GM Model for short) is generally expressed as GM(n,x) model, which means: use n-order differential equations to build a model for x variables. Grey prediction model regards discrete data scattered on the time axis as a set of continuously changing sequences, and uses the method of accumulation and subtraction to weaken the unknown factors in the grey system and strengthen the influence of known factors. Finally, a continuous differential equation with time as a variable is constructed, and the parameters in the equation are determined by mathematical methods to achieve the purpose of prediction. For example: The prediction principle of GM(1,1) model is: a set of new data sequences with obvious trends are generated for a certain data sequence by accumulation, and a model is established according to the growth trend of the new data sequence for prediction, and then the reverse calculation is performed by the method of accumulation and subtraction to restore the original data sequence, and then the prediction result is obtained.

[0082] In this scheme, based on the traditional multidimensional grey prediction model, an AGM (1, n) model is constructed to automatically screen the time window and related variables of time series data to improve the prediction accuracy of the model.

[0083] The AGM (1, n) model construction and performance comparison is based on the traditional multi-dimensional grayscale prediction model. It automatically selects the time window and related variables of the time series data to improve the prediction accuracy of the model. The night light brightness value is used to represent the intensity of regional human activity interference. The two natural elements that have the greatest impact on wetlands, temperature and precipitation, are selected to establish a prediction model for the wetland area change process based on these three factors. Then, the C value is used as the standard to compare with the traditional GM model.

[0084] Step 160: Perform spatiotemporal analysis of wetland changes based on the trend results, abnormal changes, correlation results, and disturbance factors.

[0085] When conducting spatiotemporal analysis of wetland changes, evaluation can be conducted from several dimensions, including trend results of time series predictions, abnormal changes, correlation results, and disturbance factors.

[0086] Figure 1The method in the paper obtains long-term basic data and uses a preset clustering method to calculate the emerging hotspot value of each statistical area; and determines the trend result of time series prediction based on the emerging hotspot value; calculates the outliers in the long-term wetland map, and spatially clusters and detects the significant areas in the neighborhood based on the outliers to obtain the abnormal changes; based on the geographically weighted regression model, the time label is added to analyze the correlation between the wetland change area and the night light change in different regions in the long time series, and obtains the correlation results; builds an adaptive multidimensional grayscale prediction model to determine the disturbance factors affecting the change of wetland area; and conducts spatiotemporal analysis of wetland changes based on trend results, abnormal changes, correlation results and disturbance factors. The technical solution of the present invention incorporates the time dimension into the geographically weighted regression analysis method, and optimizes the adaptive multidimensional grayscale prediction model at the same time, so as to reveal the driving mechanism of interference factors on plateau urban wetlands and construct a geographical spatiotemporal mining system; the spatiotemporal changes of wetland data are detected by performing emerging spatiotemporal hotspot analysis and local outlier analysis on long-term plateau wetlands, and the spatiotemporal geographically weighted regression analysis is used to analyze the night light data and wetland classification data to determine the correlation between interference factors and wetland area changes; finally, the adaptive multidimensional grayscale prediction model method is used to improve the model accuracy, so that the optimized AGM (1, n) model has better simulation and prediction performance.

[0087] based on Figure 1 The method, the embodiments of this specification also provide some specific implementation methods of the method, which are described below.

[0088] Before calculating the emerging hotspot value of each statistical area in the plurality of statistical areas by using a preset clustering method, the method may further include:

[0089] Determining a target statistical step length based on the long time series wetland map in the long time series basic data;

[0090] A plurality of statistical areas are determined based on the target statistical step size.

[0091] As an optional implementation, step 120 may specifically include:

[0092] A hotspot analysis clustering method is adopted to calculate the emerging hotspot value of each of the multiple statistical areas based on the preset range and the target statistical step; the emerging hotspot value of the previous period is compared with the emerging hotspot value of the second period; if the emerging hotspot value of the previous period is less than the emerging hotspot value of the second period, the result is +1; if the emerging hotspot value of the previous period is greater than the emerging hotspot value of the second period, the result is -1; if the emerging hotspot value of the previous period is equal to the emerging hotspot value of the second period, the result remains unchanged, and after calculating the data of all time periods in each statistical area, a final result value is generated; the emerging hotspot value of each time period is integrated with the MannKendall trend result to obtain the trend result of time series prediction.

[0093] More specifically, for long-term wetland maps, determine the appropriate statistical step size and use GetisOrd Statistics clustering method, calculate each statistical area values, individual high-value elements may not be statistically significant hot spots.

[0094] When a high-value element and its adjacent areas are all high values, they are emerging hot spots. Similarly, when a low-value area and its adjacent areas are all low values, they are cold spots. The specific calculation is as follows: Formulas (1), (2), and (3).

[0095]

[0096] Then, the The value is integrated with the MannKendall trend result to obtain the trend result of time series prediction. i,j is the spatial weight between regions i and j, n is the number of regions, x j represents the wetland area in region j, represents the mean value of wetland area, and S represents the standard deviation of wetland area.

[0097] As an optional implementation, step 130 may include:

[0098] The z value is calculated spatially using Local Moran's I; the significant areas in the neighborhood are spatially clustered and anomaly detected based on the z value to generate clustering results; and the abnormal changes are tested on the basis of the time series according to the clustering results.

[0099] Furthermore, based on the z-value, the significant areas in the neighborhood are spatially clustered and anomaly detected to generate a clustering result, including: for any of the statistical areas, if the z-value of the statistical area is a positive value higher than the first preset threshold, it means that the areas around the statistical area are similar values; if the areas around the statistical area are all high values, the clustering result is a high-high cluster; if the areas around the statistical area are all low values, the clustering result is a low-low cluster; if the z-value of the statistical area is a negative value lower than the second preset threshold, it means that there is a large difference between the statistical area and the surrounding areas, which is a statistically significant spatial data outlier; the second preset threshold is less than the first preset threshold.

[0100] In specific implementation, for example, Local Moran's I can be used to calculate the z value in space, perform spatial clustering and anomaly detection on the significant areas in the neighborhood, and generate clustering results. If the z value of a region is a high positive value, it means that the surrounding areas have similar values ​​(high or low values). If the surrounding areas are all high values, it is HH (high-high clustering); if the surrounding areas are all low values, it is LL (low-low clustering). If the z value in a region is a low negative value, it means that there is a large difference between the region and the surrounding areas, and it is a statistically significant spatial data outlier. The specific calculation formulas are shown in (4), (5), (6), (7), and (8). The abnormal changes are tested based on the time series.

[0101]

[0102] V[I i ]=E[I i 2 ]-E[I i ] 2 (8)

[0103] Among them, x i is the wetland area of ​​region i, is the average wetland area, w i,j is the spatial weight between regions i and j, n is the number of regions, I i represents the Moran index of region i, represents the sample standard deviation of wetland area, represents the z-statistic score of the Moran index of region i, E[I i ] indicates I i The expected value of V[I i ] indicates I i The variance of .

[0104] Step 140 may include:

[0105] A geographically weighted regression model is used to model the change in wetland area and the intensity of night light changes in a preset area, and the results are classified according to the correlation coefficient to obtain classification results; the classification results are used to indicate the degree of influence of disturbance factors on wetland area changes in different regional areas.

[0106] More specifically, based on the Geographically Weighted Regression model (GWR), adding time tags can not only reflect the spatial location characteristics of the model but also take into account the time factor, which can effectively help analyze the spatiotemporal characteristics of the data. The correlation between the change in wetland area and night light changes in different regions over a long time series is analyzed.

[0107] When implemented, each position (u i ,v i The SGWR calculation method of ) is shown in formula (9):

[0108] y ti =β0(u i ,v i )+β1(u i ,v i )x ti +ε i (9)

[0109] Among them, y ti represents the time variation coefficient of the dependent variable at the ith location, i.e., the wetland area change, (u i ,v i ) represents the coordinates of the i-th position, β0(u i ,v i ) represents the intercept at the i-th position, β1(u i ,v i ) represents the regression parameter of the i-th position, x ti represents the time variation coefficient of the independent variable at the ith position, i.e., the nighttime light value, ε i represents the random error term. The parameter estimates are calculated as follows:

[0110]

[0111] Among them, W represents the spatial weight matrix. The impact on wetland area change mainly depends on the regional effect within the neighborhood, and the impact factors of other regions are relatively low. Therefore, bisquare is used as the kernel function:

[0112]

[0113] y in formula (9) ti and x tiis the time variation coefficient of the corresponding variable, and the trend line is represented by a linear equation. Its regression coefficient β1 is taken as the time variation coefficient. The calculation method is as follows:

[0114] y i =β0+β1x i +ε i (12)

[0115] Among them, y i Indicates wetland area, night light value, x i represents the sample time, β1 represents the regression coefficient, and the least squares method is used for calculation.

[0116] Optionally, step 150 may include:

[0117] Select a time window; calculate the target coefficient of the adaptive multidimensional grayscale prediction model in combination with the relevant variables in each time window; the adaptive multidimensional grayscale prediction model is a model trained based on the long-term wetland map, the long-term nighttime light satellite data, the long-term annual precipitation data and the annual temperature and climate data; select the time window and related variables corresponding to the minimum target coefficient; determine the target key climate factors and disturbance factors affecting the wetland area in the corresponding area.

[0118] AGM (1,n) model construction and performance comparison: Based on the traditional multidimensional gray prediction model, the time window and related variables of the time series data are automatically screened to improve the prediction accuracy of the model. The night light brightness value is used to represent the intensity of regional human activity interference, and the two natural factors that have the greatest impact on wetlands - temperature and precipitation - are selected. The prediction model for the wetland area change process based on these three factors is established.

[0119] Next, in order to explain the technical solution provided by the present invention in detail, the implementation process and technical effects of the present invention are described by way of examples:

[0120] For example, taking the plateau urban wetlands in Yunnan Province as an example to analyze the mechanism of long-term human disturbance:

[0121] First, data preparation is carried out: wetland data, night light data, temperature and precipitation data can be collected and processed.

[0122] When collecting wetland data, for example: for experimental operation, in actual operation, the wetland data is from Zhao (Zhao, F., Feng, S., Xie, F., et al., 2023. Extraction of long time series wetland information based on Google Earth Engine and random forest algorithm for a plateau lake basin-a case study of Dianchi Lake, Yunnan Province, China. Ecol. Indic. 146, 109813.) et al. The long time series wetland map extracted from Landsat images by random forest classification method, the time range is 1993-2020, the spatial resolution is 30m; the overall accuracy of the data is above 80%, and the quantitative and allocation divergence index (QADI) is not higher than 0.2. These data were used to extract and analyze the area, distribution characteristics, structure and spatiotemporal change characteristics of the wetlands in the Dianchi Lake Basin.

[0123] When collecting night light data, for example, for experimental operations, in actual operations, the night light data comes from the calibration data of the Defense Meteorological Satellite Program (DMSP) / Operational Linescan System (OLS) and the Visible Infrared Imaging Radiometer Suite (VIIRS) coordinated by Li et al. (Li, X., Zhou, Y., Zhao, M., et al., 2020. A harmonized global nighttime light dataset 1992-2018. Sci. Data 7(1), 1-9.). This data is a fusion of night light data taken by two different satellites to generate a consistent long-term night light data with a spatial resolution of 1 km and a time period of 1993-2020.

[0124] The precipitation and temperature data are based on the global 0.5° climate dataset released by the Climate Research Center (CRU) of the University of Anglia, UK, and the global high-resolution climate dataset released by WorldClim. The dataset is generated in China through the Delta spatial downscaling scheme. The dataset is interannually synthesized to obtain precipitation and temperature data in the corresponding area. The spatial resolution is 0.0083333°, about 1km. The time range is from January 1993 to December 2020. The data format is NetCDF (Network Common data Form) multidimensional data format. The data unit is 0.1mm. This is a dataset generated in China through the Delta spatial downscaling scheme based on the global 0.5° climate dataset released by the Climate Research Center (CRU) of the University of Anglia, UK, and the global high-resolution climate dataset released by WorldClim. The monthly precipitation data and monthly average temperature data in the study area are interannually synthesized and used as interannual data for further research.

[0125] When implementing this solution based on the data collected above, the corresponding implementation process may include:

[0126] 1) Calculation of new hotspots: The emerging hotspot analysis method was used to calculate the wetland change area of ​​Dianchi Lake from 1993 to 2020 with a 2km neighborhood and a 3-year time step. The values ​​are tested by Mann-Kendall trend test to evaluate the development trend of hot spots and cold spots. The value of the second cycle The result is +1 when the former is less than the latter, -1 when the former is greater than the latter, and unchanged when the former is equal to the latter. After calculating the data of all time periods in each region, the final result value is generated. The values ​​are integrated with the MannKendall trend results to obtain the trend results of time series prediction. Among them, the definition of New Hot Spot means that it was a hot spot last time, but not before; continuous hot spot means that up to 90% of the time steps are hot spots without interruption; persistent hot spot means that the hot spot appears in more than 90% of the time steps; sporadic hot spot means that the hot spot appears repeatedly in history, but not continuously. On this basis, the time series trend results of wetlands are obtained.

[0127] 2) Calculation of outliers: With a 2km neighborhood and a 3-year time step, after Local Moran's I calculation and classification of each period of data on the time step, the final spatiotemporal local anomaly detection category is generated. This manual conducts local outlier analysis on the spatiotemporal changes in wetland area, and the p value is set to 0.05. If the area of ​​each period of data is HH, it is Only High-HighCluster; if the area of ​​each period of data is LL, it is Only Low-Low Cluster; if the area of ​​each period of data is HL, it is Only High-Low Outlier; if the area of ​​each period of data is LH, it is Only Low-HighOutlier; if the regional clustering type of each period of data is not fixed, it is Multiple Types. And use Mann-Kendall to perform trend prediction for the classification results within each step to test the abnormal changes based on the time series.

[0128] 3) Calculation of driving force of disturbance factors (taking human disturbance as an example): This paper uses SGWR to model the wetland area and NTL intensity in the Dianchi Lake Basin, and classifies the results according to the correlation coefficient to clarify the degree of influence of human activity intensity in different regions on the change of wetland area. This method changes the limitations of traditional geographically weighted regression analysis in terms of time scale, innovatively combines the temporal change trend of target features with spatial location, and conducts spatiotemporal geographically weighted regression analysis with geographical significance.

[0129] 4) Model construction and performance comparison: This specification proposes an adaptive multidimensional grey prediction model, which automatically selects the time window and related variables of the time series data of each region, thereby improving the prediction accuracy of the model. Figure 2 As shown, the time window starts from 1993 and is delayed by one year each time. The time window selected in this specification includes at least 3 years, namely 1993-2020, 1994-2020, ... and 2018-2020. Combined with the variables in each time window, the C of the grey prediction model is calculated, the time window and related variables when C is minimum are selected, and Eq. (19) is calculated. This method calculates 5 wetland areas divided into different levels according to the degree of human activity impact through AGM (1, n), and determines the corresponding prediction model parameters from the adaptively selected indicators to obtain the most critical climatic factors and human factors affecting the wetland area in the corresponding area. This method cuts the time windows and variables separately and combines them with each other, adaptively selects the best time windows and variables to obtain a higher precision model, and provides a unique innovative perspective for realizing high-precision time prediction models and screening important driving forces at a macro scale.

[0130] 5) Changes and analysis of spatiotemporal hotspots in the wetlands of the Dianchi Lake Basin: The hotspots are mainly concentrated around Dianchi Lake, with a few distributed in the northeast of Dianchi Lake. The hotspots around Dianchi Lake are mainly continuous hotspots and persistent hotspots, and the areas expanding outward are mainly new hotspots. The above phenomenon shows that the wetlands in the north, east and south of Dianchi Lake existed all year round from 1993 to 2020 and were maintained well without large-scale damage. In addition, new hotspots have appeared in the outer circles of the north and west of Dianchi Lake, indicating that wetland protection policies have achieved remarkable results in recent years, and the wetland area in the Dianchi Lake Basin has increased significantly, mainly concentrated in the north and west of Dianchi Lake. Figure 3 As shown, each cube represents a wetland hotspot for a time step of 3 years. It can be seen that some areas around Dianchi Lake have been hotspots for many years with a confidence level of 90%, while the northern and eastern areas of Dianchi Lake are new hotspots in recent years. From 1993 to 2020, the north and south banks of Dianchi Lake have long been hotspots for wetland development, while in the past 3-6 years, the east and north banks of Dianchi Lake have become new areas with active wetland development, indicating that the wetland protection policy in the Dianchi Basin has a clear directional development.

[0131] 6) Analysis of local spatiotemporal changes and abnormal changes in wetlands in the Dianchi Lake Basin: The results are as follows: Figure 4 As shown, the periphery of the Dianchi Lake basin is dominated by Low-Low clusters. The north, east and west of Dianchi Lake are dominated by High-High clusters, with the largest number in the north, extending 20.02km to the north of Dianchi Lake. In addition, some Low-High clusters are scattered in the interior of Dianchi Lake and some areas in the east and south, which are always low-density areas surrounded by high wetland density areas. However, there are some multi-type areas in the north and south of Dianchi Lake, and the cluster types in these areas are not fixed between 1993 and 2020, and have undergone certain changes. In the periphery of the Dianchi Lake basin, Low-Low clusters are the main areas, indicating that there are wetlands in this area, but at a lower level compared with the surrounding areas of Dianchi Lake. Each small cube represents an abnormal change in a time step, and the time step is 3 years. Specifically, the multi-type area is mainly a mixed area of ​​low-high clusters and high-high clusters. The three areas in the north of Dianchi Lake were mixed with High-High clusters and Low-High clusters for a period of time, indicating that the wetland area in this area was smaller than the surrounding areas in the corresponding period (2005-2008), and became a low wetland density area surrounded by high wetland density. The multi-type area on the south side of Dianchi Lake changed from a High-High agglomeration area to a Low-High agglomeration area or an insignificant agglomeration area. There are two possibilities for this situation: the surrounding wetland area increases rapidly, or the wetland area in the area decreases. By observing the changes in the wetland area in the area and its surrounding areas, it was found that the main reason was that the wetland density in the area was once very high, but in the subsequent period of time, due to the influence of human or natural factors, the wetland density continued to decline, and even the wetlands disappeared completely.

[0132] 7) Calculation and analysis of human disturbance in the Dianchi Lake Basin: The human disturbance in the Dianchi Lake Basin was calculated by SGWR. The correlation between wetland area change and NTL change was divided into five levels according to the natural breakpoint method, namely strong negative correlation, weak negative correlation, weak correlation, weak positive correlation, and strong positive correlation. Figure 5 As shown. More specifically, the relationship between the change in wetland area and the change in nighttime light is mainly weak negative correlation. There is a weak positive correlation distribution area on the west side of Dianchi Lake, and there are weak positive correlation and strong positive correlation distribution areas on the south side of Dianchi Lake. The situation in the northern part of Dianchi Lake is complicated, with positive correlation areas and negative correlation areas alternating, indicating that the human interference factors for wetland changes in this area are large and inconsistent. Among them, the 4.25km range in the north of Dianchi Lake is a weak negative correlation area, the northwest of Dianchi Lake is a negative correlation area and a positive correlation area respectively, and the center is a strong negative correlation area and a strong positive correlation area. Further north is an area centered on weak negative correlation, and negative and positive correlations are symmetrically distributed. In other areas of the Dianchi Lake basin, there are also some areas disturbed by human activities. In general, the areas with greater human interference in the Dianchi Lake basin are mainly concentrated around Dianchi Lake and the north side of Dianchi Lake. The human interference phenomenon on the north side of Dianchi Lake is more complicated. Human activities are negatively correlated with wetland changes, such as Figure 6 As shown in the figure, the wetland area has decreased under the negative influence of human interference. It can be seen that the wetlands in the Dianchi Lake Basin are mainly concentrated around Dianchi Lake, with good overall protection and continuous expansion to the surrounding areas. Unfortunately, due to uneven protection, the speed of change in the wetland area in the north and south of Dianchi Lake does not match that of the surrounding areas, and has even decreased.

[0133] 8) Results and performance comparison of the AGM (1, n) prediction model in the Dianchi Lake Basin: The changes in the spatiotemporal hot spots of the wetlands in the Dianchi Lake Basin and the local spatiotemporal abnormal changes further reveal the reasons behind the results. Therefore, based on the calculation of human disturbance and combined with natural factors, the AGM (1, n) model is used to fit the prediction. Using the dynamic sliding time window, we can find the time series data that best suits the current trend, automatically select the dependent variable, and establish an accurate AGM (1, n) model. If there is a strong negative correlation in the area, NTL is not used as the dependent variable. This is because although human disturbance has a great impact on the changes in wetlands in the region, compared with other factors, changes in natural climate are more consistent with changes in wetland area. The final selected time window, dependent variable combination, and posterior odds ratio are shown in Table 2. The longer the time series, the higher the prediction accuracy, and the main time window is concentrated in the last 4-5 years. When the trend of changes in independent and dependent variables can be tracked, a longer time window will provide more accurate model parameters. However, when the data in the subsequent time series and the data in the previous time series mutate due to natural changes, severe human interference and other factors, the previous historical data is an interference factor for the model prediction, and the time window of the most recent period can provide more accurate model parameters than the overall time window. In addition, it can be seen from the results that the temperature factor has a considerable impact on each region, followed by the intensity of human activities, and finally the precipitation factor. This shows that the dominant factor affecting the change of wetland area in the Dianchi Lake Basin is temperature, and the impact of human activity intensity on the change of wetland area in some parts of the basin is more significant than the impact of precipitation. The comparison of the C value between the improved AGM (1, n) model and the (1, 1) model is shown in Table 2. In summary, the improved AGM (1, n) in this manual has high availability in dependent variable selection and prediction accuracy.

[0134] Table 1 AGM (1, n) fitting results

[0135]

[0136] Table 2 Model comparison results

[0137]

[0138] From the above examples, we can see that the present invention combines long-term wetland data, NTL human disturbance data, natural factor data with improved SGWR and AGM algorithms to conduct spatiotemporal analysis of plateau wetlands. It demonstrates the feasibility of multi-factor combination and the superiority of optimization algorithms, better reproduces the spatiotemporal evolution process of plateau wetlands and their driving conditions, and can provide data support and scientific basis for the sustainable management of plateau wetlands. This method can be used for the spatiotemporal analysis of human disturbance mechanisms in other long-term wetland series.

[0139] Therefore, the present invention incorporates the time dimension into the geographically weighted regression analysis method, optimizes the adaptive multidimensional grayscale prediction model, reveals the driving mechanism of human disturbance on plateau urban wetlands, and constructs a geographical spatiotemporal mining system. The spatiotemporal changes of wetland data are detected by performing emerging spatiotemporal hot spot analysis (Emerging Hot Spot Analysis) and local outlier analysis (Local Outlier Analysis) on long-term plateau wetlands. Then, the spatiotemporal geographically weighted regression analysis (Spatiotemporal Geographically Weighted Regression, SGWR) is used to analyze the nighttime light data (Nighttime light, NTL) and wetland classification data to explore the correlation between human disturbance and wetland area changes. Finally, the adaptive multidimensional grayscale prediction model (AGM (1, n)) method is used to establish a wetland area prediction model using NTL, annual temperature data and annual precipitation data, which can improve the model accuracy and make the optimized AGM (1, n) model have better simulation and prediction performance.

[0140] Based on the same idea, the present invention also provides a spatiotemporal analysis device for plateau wetland disturbance mechanism, such as Figure 8 As shown, the device may include:

[0141] The long-term basic data acquisition module 810 is used to acquire long-term basic data; the long-term basic data at least includes a long-term wetland map, long-term nighttime light satellite data, long-term annual precipitation data, and annual temperature and climate data;

[0142] The emerging hotspot value calculation module 820 is used to calculate the emerging hotspot value of each statistical area by using a preset clustering method; and determine the trend result of the time series prediction based on the emerging hotspot value; the statistical area is determined based on the long time series wetland map;

[0143] An outlier calculation module 830 is used to calculate outliers in the long-term wetland map, and perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes;

[0144] The correlation analysis module 840 is used to analyze the correlation between the wetland change area and night light change in different regions of the long time series based on the geographical weighted regression model and adding time tags to obtain the correlation results;

[0145] An adaptive multi-dimensional grayscale prediction model building module 850 is used to build an adaptive multi-dimensional grayscale prediction model to determine disturbance factors that affect wetland area changes;

[0146] The spatiotemporal analysis module 860 is used to perform spatiotemporal analysis on wetland changes based on the trend results, abnormal changes, correlation results and disturbance factors.

[0147] based on Figure 8 The device in the embodiment may further include some specific implementation units:

[0148] Optionally, the device may further include:

[0149] A target statistical step length determination module, used to determine the target statistical step length based on the long time series wetland map in the long time series basic data;

[0150] The statistical region determination module is used to determine a plurality of statistical regions based on the target statistical step length.

[0151] Optionally, the emerging hotspot value calculation module 820 may include:

[0152] An emerging hotspot value calculation unit, used to calculate the emerging hotspot value of each of the plurality of statistical areas based on a preset range and the target statistical step length by using a hotspot analysis clustering method;

[0153] The emerging hotspot value comparison unit is used to compare the emerging hotspot value of the previous cycle with the emerging hotspot value of the second cycle; if the emerging hotspot value of the previous cycle is less than the emerging hotspot value of the second cycle, the result is +1; if the emerging hotspot value of the previous cycle is greater than the emerging hotspot value of the second cycle, the result is -1; if the emerging hotspot value of the previous cycle is equal to the emerging hotspot value of the second cycle, the result remains unchanged, and after calculating the data of all time periods in each of the statistical areas, a final result value is generated;

[0154] The trend result determination unit is used to integrate the emerging hotspot values ​​of each time period with the MannKendall trend results to obtain the trend results of time series prediction.

[0155] Optionally, the outlier calculation module 830 may include:

[0156] The z-value calculation unit is used to calculate the z-value spatially using Local Moran's I;

[0157] A clustering result generating unit, used for performing spatial clustering and anomaly detection on the significant areas in the neighborhood based on the z value, and generating clustering results;

[0158] The abnormal change checking unit is used to check the abnormal change based on the time series according to the clustering result.

[0159] Optionally, the clustering result generating unit may be used to:

[0160] For any of the statistical regions, if the z value of the statistical region is a positive value higher than the first preset threshold, it means that the regions around the statistical region are similar values; if the regions around the statistical region are all high values, the clustering result is a high-high cluster; if the regions around the statistical region are all low values, the clustering result is a low-low cluster;

[0161] If the z value of the statistical area is a negative value lower than the second preset threshold, it means that there is a large difference between the statistical area and the surrounding area, and it is a statistically significant spatial data outlier; the second preset threshold is smaller than the first preset threshold.

[0162] Optionally, the correlation analysis module 840 may be used to:

[0163] A geographically weighted regression model is used to model the change in wetland area and the intensity of night light changes in a preset area, and the results are classified according to the correlation coefficient to obtain classification results; the classification results are used to indicate the degree of influence of disturbance factors on wetland area changes in different regional areas.

[0164] Optionally, the adaptive multi-dimensional grayscale prediction model building module 850 may include:

[0165] A time window selection unit, used for selecting a time window;

[0166] A target coefficient calculation unit is used to calculate the target coefficient of the adaptive multidimensional grayscale prediction model in combination with relevant variables in each time window; the adaptive multidimensional grayscale prediction model is a model trained based on the long-term wetland map, the long-term nighttime light satellite data, the long-term annual precipitation data and the annual temperature and climate data;

[0167] A time window and related variable selection unit, used to select the time window and related variables corresponding to the minimum target coefficient;

[0168] The target key climatic factors and disturbance factors determination unit is used to determine the target key climatic factors and disturbance factors that affect the wetland area in the corresponding area.

[0169] Based on the same idea, the embodiment of this specification also provides a spatiotemporal analysis device for the disturbance mechanism of plateau wetlands. Fig. 9 As shown, the device includes:

[0170] A memory, a processor and a communication interface coupled to the processor; the memory stores a computer program that can be run by the processor; when the processor runs the computer program, it executes the aforementioned spatiotemporal analysis method of a plateau wetland disturbance mechanism.

[0171] like Fig. 9As shown, the processor may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present invention. The communication interface may be one or more. The communication interface may use any transceiver-like device for communicating with other devices or communication networks.

[0172] like Fig. 9 As shown, the terminal device may further include a communication line. The communication line may include a path to transmit information between the components.

[0173] Optional, such as Fig. 9 As shown, the terminal device may further include a memory. The memory stores a computer program that can be executed by the processor; when the processor executes the computer program, the method provided by the embodiment of the present invention is implemented.

[0174] Optionally, the computer-executable instructions in the embodiment of the present invention may also be referred to as application program codes, which is not specifically limited in the embodiment of the present invention.

[0175] In a specific implementation, as an example, Fig. 9 As shown, the processor may include one or more CPUs, such as Fig. 9 CPU0 and CPU1 in.

[0176] In a specific implementation, as an example, Fig. 9 As shown, the terminal device may include multiple processors, such as Fig. 9 Each of these processors can be a single-core processor or a multi-core processor.

[0177] Based on the same idea, the embodiments of this specification also provide a computer storage medium corresponding to the above embodiments, and the computer storage medium stores instructions, and when the instructions are executed, the method in the above embodiments is implemented.

[0178] The above mainly introduces the solution provided by the embodiment of the present invention from the perspective of the interaction between the various modules. It is understandable that, in order to realize the above functions, each module includes a hardware structure and / or software unit corresponding to the execution of each function. It should be easily appreciated by those skilled in the art that, in combination with the units and algorithm steps of each example described in the embodiments disclosed in this specification, the present invention can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0179] The embodiment of the present invention can divide the functional modules according to the above method example. For example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiment of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0180] The processor in this specification may also have the function of a memory. The memory is used to store computer-executable instructions for executing the solution of the present invention, and the execution is controlled by the processor. The processor is used to execute the computer-executable instructions stored in the memory, thereby implementing the method provided by the embodiment of the present invention.

[0181] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art may understand and implement other variations of the disclosed embodiments by viewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "one" or "an" does not exclude multiple situations. A single processor or other unit may implement several functions listed in the claims. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

[0182] Although the present invention has been described in conjunction with specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present invention. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present invention as defined by the appended claims and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present invention. Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, the present invention is intended to include such modifications and variations if they fall within the scope of the claims of the present invention and their equivalents.

Claims

1. A spatiotemporal analysis method for plateau wetland disturbance mechanisms, characterized in that: Methods include: Acquire long-term basic data; the long-term basic data at least includes long-term wetland maps, long-term nighttime light satellite data, long-term annual precipitation data, and annual temperature and climate data; The preset clustering method is used to calculate the emerging hotspot value for each statistical area; and determining the trend result of the time series prediction based on the emerging hotspot value; the statistical area is determined based on the long-term wetland map; Calculate the outliers in the long-term wetland map, and perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes; Based on the geographically weighted regression model, the time label is added to analyze the correlation between the wetland change area and night light changes in different regions of the long time series, and obtain the correlation results; Build an adaptive multi-dimensional grayscale prediction model to determine the disturbance factors that affect wetland area changes; A spatiotemporal analysis of wetland changes was conducted based on the trend results, abnormal changes, correlation results, and disturbance factors.

2. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 1 is characterized in that: Before calculating the emerging hotspot value of each statistical area in the plurality of statistical areas by using a preset clustering method, the method further includes: Determining a target statistical step length based on the long time series wetland map in the long time series basic data; A plurality of statistical areas are determined based on the target statistical step size.

3. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 2, characterized in that: The method of using a preset clustering method to calculate the emerging hotspot value of each statistical area in the plurality of statistical areas; And determine the trend results of time series forecasting based on emerging hotspot values, including: Using a hotspot analysis clustering method to calculate the emerging hotspot value of each of the plurality of statistical areas based on a preset range and the target statistical step; Compare the emerging hotspot values ​​of the previous cycle with the emerging hotspot values ​​of the second cycle; If the value of the emerging hotspot in the previous cycle is less than the value of the emerging hotspot in the second cycle, the result is +1; If the value of the emerging hotspot in the previous cycle is greater than the value of the emerging hotspot in the second cycle, the result is -1; If the value of the emerging hotspot in the previous cycle is equal to the value of the emerging hotspot in the second cycle, the result remains unchanged, and the final result value is generated after calculating the data of all time periods in each of the statistical areas; The emerging hotspot values ​​of each time period are integrated with the MannKendall trend results to obtain the trend results of time series forecasting.

4. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 1, characterized in that: The calculating of the outliers in the long-term wetland map and the spatial clustering and anomaly detection of the significant areas in the neighborhood based on the outliers to obtain the abnormal changes include: Use Local Moran's I to calculate z-values ​​spatially; Based on the z value, the significant areas in the neighborhood are spatially clustered and anomaly detected to generate clustering results; According to the clustering results, abnormal changes are tested on the basis of time series.

5. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 4, characterized in that: Based on the z value, the significant areas in the neighborhood are spatially clustered and anomaly detected to generate clustering results, including: For any of the statistical regions, if the z value of the statistical region is a positive value higher than the first preset threshold, it means that the regions around the statistical region are similar values; if the regions around the statistical region are all high values, the clustering result is a high-high cluster; if the regions around the statistical region are all low values, the clustering result is a low-low cluster; If the z value of the statistical area is a negative value lower than the second preset threshold, it means that there is a large difference between the statistical area and the surrounding area, and it is a statistically significant spatial data outlier; the second preset threshold is smaller than the first preset threshold.

6. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 1, characterized in that: Based on the geographically weighted regression model, the time label is added to analyze the correlation between the wetland change area and night light change in different regions of the long time series, and the correlation results are obtained, including: A geographically weighted regression model is used to model the change in wetland area and the intensity of night light changes in a preset area, and the results are classified according to the correlation coefficient to obtain classification results; the classification results are used to indicate the degree of influence of disturbance factors on wetland area changes in different regional areas.

7. The spatiotemporal analysis method of plateau wetland disturbance mechanism according to claim 1, characterized in that: The adaptive multi-dimensional grayscale prediction model is constructed to determine the disturbance factors affecting the change of wetland area, including: Select the time window; In combination with the relevant variables in each time window, the target coefficient of the adaptive multidimensional grayscale prediction model is calculated; the adaptive multidimensional grayscale prediction model is a model trained based on the long-term wetland map, the long-term nighttime light satellite data, the long-term annual precipitation data and the annual temperature and climate data; Select the time window and related variables corresponding to the minimum target coefficient; Determine the target key climatic factors and disturbance factors that affect the wetland area in the corresponding region.

8. A spatiotemporal analysis device for plateau wetland disturbance mechanisms, characterized in that: The device includes: A long-term basic data acquisition module is used to acquire long-term basic data; the long-term basic data at least includes a long-term wetland map, long-term nighttime light satellite data, long-term annual precipitation data, and annual temperature and climate data; An emerging hotspot value calculation module is used to calculate the emerging hotspot value of each statistical area using a preset clustering method; and determining the trend result of the time series prediction based on the emerging hotspot value; the statistical area is determined based on the long-term wetland map; An outlier calculation module is used to calculate outliers in the long-term wetland map, and to perform spatial clustering and anomaly detection on significant areas in the neighborhood based on the outliers to obtain abnormal changes; A correlation analysis module is used to analyze the correlation between the wetland change area and night light change in different regions of the long time series based on a geographically weighted regression model and adding a time tag to obtain a correlation result; Adaptive multi-dimensional grayscale prediction model building module, used to build an adaptive multi-dimensional grayscale prediction model to determine the disturbance factors that affect the change of wetland area; The spatiotemporal analysis module is used to perform spatiotemporal analysis on wetland changes based on the trend results, abnormal changes, correlation results and disturbance factors.

9. A device for spatiotemporal analysis of plateau wetland disturbance mechanisms, characterized in that the device include: A memory, a processor, and a communication interface coupled to the processor; The memory stores a computer program executable by the processor; When the processor runs the computer program, it executes the spatiotemporal analysis method of the plateau wetland disturbance mechanism as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: The computer storage medium stores instructions, and when the instructions are executed, the spatiotemporal analysis method of the plateau wetland disturbance mechanism described in any one of claims 1 to 7 is implemented.

Images