Long-time-sequence high-frequency ecological environment quality space-time differentiation and driving analysis method

Abstract

The invention specifically discloses a long-time-sequence high-frequency ecological environment quality space-time differentiation and driving analysis method, and relates to the technical field of remote sensing and ecological environment. The method comprises the following steps: determining an evaluation area, and constructing a macroscopic ecological safety risk evaluation framework; ecological indexes of greenness, humidity, temperature and dryness are calculated, and a time sequence data set is constructed; reconstructing a time sequence data set and constructing a remote sensing ecological index model based on the reconstructed time sequence data set; verifying the precision of the model, and evaluating the reconstruction precision by taking the screened high-quality pixels as true values; analyzing the spatial and temporal change trend and significance output by the model by using slope estimation and trend test, outputting future change continuity by using a Hurst index analysis model, and measuring spatial autocorrelation output by the model by using a Moran index; remote sensing ecological index evolution factors are researched by means of an optimal parameter geographic detector. According to the invention, the precision and timeliness of ecological assessment are improved, and decision support is provided for ecological management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing and ecological environment, and particularly relates to a long-time-series high-frequency ecological environment quality spatio-temporal differentiation and driving force analysis method. BACKGROUND

[0002] Ecological evaluation and ecological change condition monitoring are one of the key measures to ensure national ecological security. Among them, the ecological environment index (Ecological Environment Index, EI) is one of the most widely used indicators for quantifying ecological environment quality. With the continuous development of remote sensing technology, remote sensing ecological index (Remote Sensing Ecological Index, RSEI) emerges as the times require. It replaces the statistical indicators in EI with remote sensing indicators, and selects the first principal component to represent the ecological environment quality by means of principal component analysis (Principal Component Analysis, PCA) method, which can realize rapid and objective evaluation of ecological quality, and has been widely concerned in the field of ecological environment monitoring.

[0003] However, in the process of using remote sensing data for statistical analysis, due to the limitation of remote sensing transmission conditions, it is difficult to effectively monitor the regional ecological quality in long time series, high frequency and multi-scale. Specifically, when using the current RSEI model for monitoring, the summer with lush vegetation is usually chosen for evaluation, but this period is greatly affected by meteorological factors, and it is often difficult to obtain a sufficient number of sunny and cloudless remote sensing images within the monitoring period, which seriously affects the data integrity and effectiveness. This problem makes the constructed remote sensing data sequence have a large number of missing values and outliers, which not only reduces the accuracy and reliability of the RSEI model calculation results, but also cannot accurately reflect the long-term dynamic change characteristics of regional ecological environment quality, thereby restricting the in-depth study of the ecological environment change trend and driving factors, and making it difficult to meet the needs of macro ecological environment management and decision-making. SUMMARY

[0004] The purpose of the present application is to provide a long-time-series high-frequency ecological environment quality spatio-temporal differentiation and driving force analysis method, which effectively makes up for the defect of unstable time series of remote sensing data in the RSEI model based on the HANTS algorithm, realizes the evaluation and monitoring of regional ecological quality in long time series, high frequency and multi-dimension, objectively quantifies the ecological change trend and evolution pattern, and deeply studies the influencing factors, thereby providing scientific and accurate technical support for ecological environment management, ecological protection planning and ecological security barrier construction.

[0005] To achieve the above purpose, the present application provides a long-time-series high-frequency ecological environment quality spatio-temporal differentiation and driving force analysis method, and the specific steps are as follows: Step S1, determining the ecological quality assessment region range based on climate, temperature, topography, geology and ecosystem landscape pattern difference, and constructing a macro ecological security risk assessment framework; Step S2, calculating four ecological indexes of greenness, wetness, temperature and dryness based on MODIS image data, and constructing a time series dataset based on the four ecological indexes; Step S3, using time series harmonic method HANTS reconstructing the time series dataset to eliminate the data missing and instability in time, and obtaining the reconstructed time series dataset; Step S4, based on the reconstructed time series dataset, performing global normalization, and using principal component analysis PCA selecting the first principal component, using the greenness index as the control vector to select the direction as the ecological comprehensive index, selecting the multi-phase average value to represent the ecological environment quality of the four ecological indexes in the time series dataset, and constructing a remote sensing ecological index model; Step S5, verifying the accuracy of the remote sensing ecological index model, dividing the pixel data in the MODIS image data according to the quality control QA band, screening out the high-quality pixels as the true value, and comparing with the reconstructed time series dataset to evaluate the reconstruction accuracy; Step S6, outputting the time series set based on the remote sensing ecological index model, using slope estimation and trend test to analyze the spatial and temporal variation trend and significance of the time series set, using index spatial distribution to analyze the future change persistence of the time series set, and using Moran index to measure the spatial autocorrelation of the time series set; Step S7, based on the best parameter geographic detector, studying the factors of remote sensing ecological index evolution.

[0006] Preferably, in step S2, the greenness index is represented by the normalized difference vegetation index NDVI , and the calculation formula is as follows: ; The wetness index Wet is calculated by using the tasseled cap transformation, and the formula is as follows: ; wherein, , , , , , and are the surface reflectance of the bands in the multispectral image, , , , , , and These are their respective coefficients; Dryness index NDBSI From bare soil index BI and building indicators IBI The synthesis formula is as follows: ; ; ; in, and Surface reflectance in each band of the multispectral image; Temperature index is determined by surface temperature. LST The formula is as follows: ; in, For brightness, The wavelength is in the thermal infrared band. It is a physical constant. ; For the specific emissivity.

[0007] Preferably, in step S3, the time series dataset is reconstructed, and the specific steps are as follows: Using time series harmonic methods HANTS Processing time-series images of various ecological factors, from The images were processed using Fast Fourier Transform. FFT Decompose the time profile into The time series can be reconstructed from each harmonic component, using the following formula: ; ; in, , and These represent the original sequence, the reconstructed sequence, and the error sequence, respectively. The observation time of the remote sensing image. This represents the number of periods in the time series. , Indicates frequency as The coefficients of the trigonometric components at time , This is the average value of the entire sequence.

[0008] Preferably, in step S4, the reconstructed time series dataset is subjected to dimensionless processing and global normalization, as shown in the following formula: ; in, For specific ecological indicators Each pixel in time t Dimensionless results For specific ecological indicators Each pixel in time t The original data, min and max These are functions for the minimum and maximum values, respectively. The average value of multiple time phases is used as the ecological environment quality status in a specific period, and the formula is as follows: ; in, The ecological environment quality status for a specific period of time. , The observation time of the remote sensing image; , For frequency The coefficients of the trigonometric components at time; according to PCA Principal component analysis and based on the components NDVI Eigenvector value adjustment of the first principal component PC1 direction as RSEI The value is used to assess the quality of the ecological environment; based on the first principal component. NDVI The direction of the eigenvector values ​​defines the direction of the first principal component, i.e., the eigenvector. NDVI If it is the positive direction PC1 If it's the positive direction, then it's the negative direction; otherwise, it's the opposite direction. The formula is as follows: ; like RSEI direction and NDVI If the directions are opposite, the formula is as follows: ; in, Let be the direction function of the first principal component. The direction of the first principal component.

[0009] Preferably, in step S5, the remote sensing ecological index RSEI Model accuracy verification specifically involves: using the original... RSEI and the rebuilt RSEI The land cover types are categorized into five levels: excellent, good, average, poor, and very poor. RSEI Statistics on the percentage of different grades, and then the original RSEI With reconstruction RSEI By comparing the spatiotemporal distribution, a reconstruction can be performed. RSEI Accuracy verification.

[0010] Preferably, the quality control QA band divides the pixel quality into four cases: excellent quality, quality to be considered, pixels not generated due to clouds, and pixels not generated due to other factors.

[0011] Preferably, in step S6, the calculation formula of the Theil - Sen Median slope estimate is as follows: ; Where, is the trend of the ecological environment change, is the median function; Sen The significance of the slope estimate is determined by the Mann - Kendall trend test, and the formula is as follows: ; ; ; ; Where, m is the number of time series, , are respectively the data corresponding to time , ; is the sign function; Z is the result of the trend test, S is the statistic of the trend test; Hurst The exponential spatial distribution is divided into 3 cases: 0.5 < H < 1, indicating that the time series has persistence and the future trend is consistent with the past; H = 0.5 indicates that the sequence is a random sequence; 0 < H < 0.5 indicates that the sequence has anti - persistence and the future trend is opposite to the past; The spatial Moran's I index is used to evaluate the correlation degree between a specific pixel and its surrounding area, and the formula is as follows: ; Where, I is the Moran's I index, the value range is from - 1 to 1, n is the total number of pixels, is the spatial weight matrix, and are respectively the spatial attribute values of the and j - th pixels, is the average value.

[0012] Preferably, in step S7, the optimal parameter geodetector can identify the best combination of the spatial data discretization classification method and the spatial scale parameter, and accurately reveal the association between the dependent variable and the independent variable, mainly including the factor detector module and the interaction detector module; The factor detector module can detect the spatial differentiation of ecological quality and the size of the explanatory power of the influencing factors, and the formula is as follows: ; Among them, q is the size of the explanatory power, L is the division level of the driving factor, is the sample number of the hth level of the driving factor, is the total sample number of the region; And are the variance of the hth level of the dependent variable and the total variance of the region, SSW is the sum of the intra-layer variance, SST is the total variance of the region; The interaction detector module is used to identify the interaction effect between different driving factors, and to evaluate the explanatory power of the dependent variable Y, and to judge the interaction type between the two driving factors, and the specific steps are as follows: Step S71, respectively calculating the independent explanatory power of the independent variable and to the spatial differentiation of the dependent variable and ; Step S72, calculating the explanatory power when the two factors act together ; Step S73, by comparing , and the relationship between the three, determine the type of interaction.

[0013] Therefore, the present application proposes a long time series high frequency ecological environment quality spatio-temporal differentiation and driving analysis method, which has the following beneficial effects: (1) The present application uses the time series harmonic (HANTS) analysis method to reconstruct the time series, which can effectively reduce the discrete values and abnormal values in the remote sensing data, and fill in the missing values, significantly improve the stability of the RSEI model calculation process and the accuracy of the results, and enhance the robustness of the model, so that RSEI can more accurately reflect the true situation of regional ecological environment quality.

[0014] (2) The present application combines the spatial distribution characteristics of Sen slope and Hurst index to analyze the ecological environment changes in different ecological divisions, which can more intuitively and clearly show the spatio-temporal differentiation characteristics of the ecological quality of each region; with the help of Hurst index, the future change trend of the ecological environment of each region can be effectively predicted, which provides a scientific basis for formulating differentiated ecological protection strategies for different regions; at the same time, the Moran index is used to measure the spatial autocorrelation of the time series set, which can effectively quantify the aggregation or dispersion characteristics of the spatial distribution of ecological indicators, and provide key spatial correlation basis for spatio-temporal differentiation analysis.

[0015] (3) The present application utilizes the OPGD optimal parameter geographic detector model, quantitatively represents the influence degree and interaction effect of each factor on RSEI by calculating the influence of different classification methods of each driving factor on q value in numerical form; not only determines the main driving factors of ecological environment change, but also provides new technical methods and research basis for the research field of ecological environment spatial differentiation influencing factors, which helps to deeply understand the internal mechanism of ecological environment change.

[0016] The technical solutions of the present application will be further described in detail below through the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the long-time series high-frequency ecological environment quality spatial and temporal differentiation and driving analysis method of the present application; Figure 2 The annual average value of each index in the embodiment of the present application is shown in the figure; wherein, Figure 2 (a) in the figure is the annual average value of the NDVI index, Figure 2 (b) in the figure is the annual average value of the WET index, Figure 2 (c) in the figure is the annual average value of the LST index, Figure 2 (d) in the figure is the annual average value of the NDBSI index, Figure 2 (e) in the figure is the annual average value of the RSEI index; Figure 3 The RSEI grade proportion in different land cover types in July, August and September 2023 in the embodiment of the present application is shown in the figure; wherein, Figure 3 (a) in the figure is the RSEI grade proportion in different land cover types in July, Figure 3 (b) in the figure is the RSEI grade proportion in different land cover types in August, Figure 3 (c) in the figure is the RSEI grade proportion in different land cover types in September; Figure 4 The ecological environment quality status from 2000 to 2024 in the embodiment of the present application is shown in the figure; Figure 5 The index precision evaluation in the embodiment of the present application is shown in the figure; wherein, Figure 5 (a) in the figure is the index precision evaluation of NDVI, Figure 5 (b) in the figure is the index precision evaluation of WET, Figure 5 (c) in the figure is the index precision evaluation of LST, Figure 5 (d) in the figure is the index precision evaluation of NDBSI; Figure 6 The change trend of each ecological division in the embodiment of the present application is shown in the figure; wherein,Figure 6 In the figure, (a) represents the Sen slope from 2000 to 2025. Figure 6 (b) in the table represents the Hurst index from 2000 to 2025. Figure 6 (c) represents the significance level from 2000 to 2025. Figure 6 (d) represents the ecological environment quality from 2000 to 2025; Figure 7 This is a schematic diagram illustrating the transfer of different levels of RSEI in an embodiment of the present invention; Figure 8 This is a scatter plot of the global Moran's index in an embodiment of the present invention; Figure 9 This is a schematic diagram of the geographic detector results in an embodiment of the present invention; wherein, Figure 9 (a) in the image represents the results from the 2000 geospatial survey. Figure 9 (b) in the image shows the results from the 2005 geospatial survey. Figure 9 (c) in the image represents the results from the 2010 geospatial survey. Figure 9 (d) in the figure represents the results from the 2015 geospatial survey. Figure 9 (e) in the figure represents the results of the geospatial survey in 2020. Detailed Implementation

[0018] To make the technical solutions, advantages, and objectives of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below. The described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0019] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0020] Example like Figure 1 As shown, this invention provides a method for spatiotemporal differentiation and driving analysis of long-term, high-frequency ecological environment quality, with the following specific steps: Step S1: Based on differences in climate, temperature, topography, geotectonics and ecosystem landscape patterns, the scope of ecological quality assessment is determined to be 34 provincial-level administrative regions, and a macro-ecological security risk assessment framework is constructed. Step S2: Calculate four ecological indicators—greenness, humidity, temperature, and dryness—based on MODIS image data and construct a time-series dataset based on these four ecological indicators. Greenness (G) (Fig. 1), wetness (W) (Fig. 2), temperature (T) (Fig. 3), and aridity (A) (Fig. 4) of MODIS images from 2000 to 2024 were selected as ecological indicators in this example, which were calculated using MODIS image data and related formulas. NDVI Wet LST NDBSI

[0021] Greenness indicator is represented by normalized difference vegetation index (NDVI), and the calculation formula is as follows: NDVI ; Wetness indicator is represented by the formula of the Tasseled Cap transformation as follows: Wet ; wherein, , , , , , and are the surface reflectance of the bands in the multispectral image, respectively; , , , , , and are the corresponding coefficients, respectively; Aridity indicator A is synthesized by bare soil indicator B and building indicator C, and the formula is as follows: NDBSI BI IBI ; ; ; wherein, and are the surface reflectance of the bands in the multispectral image, respectively; Temperature indicator is represented by land surface temperature LST, and the formula is as follows: ; wherein, is the brightness, is the wavelength of the thermal infrared band, is a physical constant, ; is the specific emissivity.

[0022] Step S3, using time series harmonic method HANTS ​​​​​​​​​Reconstruct the time series dataset, eliminate the missing and instability of data in time, and obtain the reconstructed time series dataset, the specific steps are as follows: Use time series harmonic method HANTS Process the time series image of each ecological factor, from The fast Fourier transform is used in FFT The time profile is decomposed into The time series is reconstructed, and the formula is as follows: ; ; Among them, 、 and represent the original sequence, the reconstructed sequence and the error sequence respectively, is the observation time of remote sensing image, is the number of cycles in the time series, 、 represent the trigonometric function component coefficient when the frequency is , is the average value of the entire sequence.

[0023] Step S4, based on the reconstructed time series dataset, global normalization is carried out, and principal component analysis is used PCA Select the first principal component, use the green index as the control vector to select the direction as the ecological comprehensive index, select the multi-temporal average value to represent the ecological environment quality of the four ecological indexes in the time series dataset, and construct the remote sensing ecological index RSEI Model; Based on the reconstructed time series dataset, dimensionless treatment is carried out, global normalization is carried out, and the formula is as follows: ; Among them, is the dimensionless result of the specific ecological index pixel in time t , is the original data of the specific ecological index pixel in time t , min and max are minimum and maximum functions respectively; The multi-temporal average value is used as the ecological environment quality state of the ecological index in a specific period, and the formula is as follows: ; Among them, is the ecological environment quality state of the specific period, 、 The observation time of the remote sensing image; , For frequency The coefficients of the trigonometric components at time; according to PCA Principal component analysis and based on the components NDVI Eigenvector value adjustment of the first principal component PC1 direction as RSEI The value is used to assess the quality of the ecological environment; based on the first principal component. NDVI The direction of the eigenvector values ​​defines the direction of the first principal component, i.e., the eigenvector direction. NDVI If it is the positive direction PC1 If it's the positive direction, then it's the negative direction; otherwise, it's the opposite direction. The formula is as follows: ; like RSEI direction and NDVI If the directions are opposite, the formula is as follows: ; in, The direction function of the first principal component. The direction of the first principal component.

[0024] Step S5, Verification RSEI Model accuracy is assessed by dividing the pixel data in MODIS image data according to the quality control QA band, selecting high-quality pixels as ground truth, and comparing them with the reconstructed time series dataset. RSEI Model accuracy verification specifically involves: using the original... RSEI Output values ​​and reconstructed values RSEI Output values ​​are categorized into five levels: Excellent, Good, Average, Poor, and Very Poor, for different land cover types. RSEI Statistics on the percentage of different grades, and then the original RSEI Output values ​​and reconstruction RSEI The output values ​​are compared in terms of spatiotemporal distribution, and then reconstructed. RSEI Accuracy verification.

[0025] like Figure 2 As shown in the figure, the curves represent the average values. During the period from 2000 to 2024, greenness and humidity showed an increasing trend in most provincial-level administrative regions of the study area, while heat and dryness showed a decreasing trend, corresponding to an increase in the remote sensing ecological index. Overall, the north side showed the best performance, gradually decreasing and then increasing towards the south side. The four ecological indices are consistent with the distribution pattern of vegetation cover.

[0026] This example uses July, August, and September 2023 as cases, and will use the original RSEI Output values ​​and reconstructed values RSEIThe output values ​​are divided into 5 levels. The original index monthly composite product is constructed using monthly averages. For example... Figure 3-4 As shown, the original RSEI The output time series data may contain outliers due to unforeseen factors such as clouds and snow, thus affecting... RSEI The resulting impact, after reconstruction, can fill and repair outliers, and the details and proportions of each level in different land cover can reveal the effects of the reconstruction. RSEI The image is superior to the original. RSEI This better represents the true changes in the ecological environment. Therefore, after principal component analysis... RSEI The index compared to the original RSEI Improves robustness by reducing outliers caused by image quality issues.

[0027] The Quality Assurance (QA) band classifies pixel quality into four categories: excellent, quality under consideration, no pixels generated due to clouds, and no pixels generated due to other factors.

[0028] like Figure 5 As shown, NDVI The reconstruction effectively preserves the information of the original image and compensates for the noise problem caused by poor observation quality. The overall image value after reconstruction has a correlation of 0.972 with the "true value", showing a strong correlation. Wet , NDBSI Image reconstruction can compensate for some observation quality issues, but because HANTS is applied to long-term asymmetric sequences, it may weaken peak and trough signals, resulting in poor detail preservation. However, the correlations of 0.8035 and 0.867 both indicate strong correlations. LST During reconstruction, the reconstruction accuracy reached 0.875 with the original value, showing a strong correlation.

[0029] Step S6, based on RSEI The model outputs a time series set. The spatiotemporal variation trend and significance of the time series set are analyzed using Theil-Sen Median slope estimation and Mann-Kendall trend test. The future variation persistence of the time series set is analyzed using the exponential spatial distribution Hurst. The Moran index is used to measure the spatial autocorrelation of the time series set. The formula for calculating the Theil-Sen Median slope estimate is as follows: ; in, In view of the trend of ecological and environmental change, It is a median function; The significance of the Sen slope estimate is determined by the Mann-Kendall trend test, as shown in the following formula: ; ; ; ; where m is the number of time series, , is the time , corresponding data; is the sign function; Z is the trend test result, S is the statistic of trend test; Hurst The exponential spatial distribution is divided into three cases: 0.5<H<1, indicating that the time series has persistence, the future trend is consistent with the past; H=0.5 indicates that the sequence is a random sequence; 0<H<0.5 indicates that the sequence has anti-persistence, the future trend is opposite to the past.

[0030] As shown in Figure 8 , the spatial Moran index is used to evaluate the correlation degree of a certain pixel and the surrounding area, and the formula is as follows: ; where I is the Moran index, the value range is-1 to 1, n is the total number of pixels, is the spatial weight matrix, and are the spatial attribute values of the i-th and j-th pixels, respectively, is the average value.

[0031] As shown in Figure 6 , in the whole time period of 2000-2025, the northwest direction shows a downward trend, the southeast direction shows an upward trend, and the overall trend shows a downward trend in 2000-2009 and an upward trend in 2010-2025, and the Hurst index can better predict the change trend.

[0032] As shown in Figure 7 , each level shows a clear "adjacent transfer" trend; the retention rate of the "extreme difference" area is more than 85% in the past five periods, indicating that the worst ecological area is difficult to improve for a long time; the retention rate of the "poor" and "medium" areas fluctuates between 60% and 70%, showing that the ecological quality of the intermediate level area is more likely to have adjacent level transition; the retention rate of the "good" area is stable in the interval of 78% to 82%, indicating that most of the good ecological areas can be continuously maintained; the retention rate of the "excellent" area experiences a process of first rising, then falling, and then rising. In the overall trend, the ecological quality distribution presents the characteristics of low-level area difficult to improve, intermediate area high mobility, and high-level area relatively stable. ​

[0033] Step S7, factor research based on the optimal parameter geographical detector selection RSEI Evolutionary factors; The optimal parameter geographical detector can identify the best combination of spatial data discretization classification method and spatial scale parameter, accurately reveal the correlation between dependent variable and independent variable, mainly including factor detector module and interaction detector module; The factor detector module can detect the spatial differentiation of ecological quality and the explanatory power of influencing factors, and the formula is as follows: ; Among them, q is the explanatory power, q The value range of is [0, 1], the larger the value, the stronger the explanatory power of the driving factor on the ecological environment, L is the division level of the driving factor, is the sample number of the hth level of the driving factor, is the total sample number of the region; and are the variance of the hth level of the dependent variable and the total variance of the region, SSW is the sum of the intra-layer variance, SST is the total variance of the region; The interaction detector module is used to identify the interaction effect between different driving factors, and to evaluate the explanatory ability of the dependent variable Y, and to judge the interaction type between the two driving factors, and the specific steps are as follows: Step S71, respectively calculate the independent explanatory ability of the independent variable and to the spatial differentiation of the dependent variable and ; Step S72, calculate the explanatory ability when the two factors act together ; Step S73, by comparing the relationship between , and , determine the type of interaction.

[0034] , and The relationship between the three is shown in Table 1.

[0035] Table 1 , and relationship ;

[0036] Use OPGDThe optimal parameter geographic detector model is used to calculate the impact of different classification methods on the q-value, and the optimal parameters are selected. RSEI Variables with significant influence, analysis RSEI Factors of evolution.

[0037] Ten variables were selected: GDP, nighttime light pollution, land use type, net primary productivity, population density, elevation, slope, monthly precipitation, monthly temperature, and monthly potential evapotranspiration. All of these variables had a significant impact on RSEI. Figure 9 As shown, diagonal lines represent single-factor interactions, while others represent two-factor interactions; ★ indicates non-linear enhancement, and others indicate two-factor enhancement. In many years of single-factor detection... NPP Net primary productivity had the strongest explanatory power, followed by land use type and precipitation. In the two-factor interactions, the interaction types were both two-factor enhancement and nonlinear enhancement. NPP When land use, precipitation, and other factors interact, the interaction exhibits a two-factor enhancement pattern, meaning that the interaction between two factors is much stronger than the interaction between a single factor, and the q value shows a trend of being greater than other two-factor interactions.

[0038] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0039] Therefore, this invention provides a method for spatiotemporal differentiation and driving analysis of long-time, high-frequency ecological environment quality, through... HANTS Algorithm optimization improves the temporal stability of remote sensing data. RSEI Model accuracy, combined with zoning analysis, trend forecasting and OPGD Driven by detection, we can achieve long-term, high-frequency monitoring of ecological quality, quantify the impact of driving factors, and provide precise technical support and scientific basis for ecological management, protection planning, and the construction of security barriers.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A long time sequence high frequency ecological environment quality spatio-temporal differentiation and driving force analysis method, characterized in that, The specific steps are as follows: Step S1, determine the ecological quality assessment area range based on climate, temperature, topography, geology and ecosystem landscape pattern difference, and construct a macro ecological safety risk assessment framework; Step S2, calculate four ecological indexes of greenness, wetness, temperature and dryness based on MODIS image data, and construct a time series dataset based on the four ecological indexes; Step S3, using time series harmonic method HANTS The time series data set is reconstructed to eliminate the missing and instability of data in time, and a reconstructed time series data set is obtained. Step S4, global normalization is performed based on the reconstructed time series dataset, and principal component analysis is used PCA The first principal component is selected, the greenness index is used as the control vector, the direction is selected as the ecological comprehensive index, the multi-time average value is selected to represent the ecological environment quality of the four ecological indexes in the time series dataset, and a remote sensing ecological index model is constructed. Step S5, verify the accuracy of the remote sensing ecological index model, divide the pixel data in the MODIS image data according to the quality control QA band, select the high-quality pixels as the true value, and compare them with the reconstructed time series dataset to evaluate the reconstruction accuracy; Step S6, based on the remote sensing ecological index model output time series set, use slope estimation and trend test to analyze the spatial and temporal variation trend and significance of the time series set, use exponential spatial distribution to analyze the future change persistence of the time series set, and use Moran's index to measure the spatial autocorrelation of the time series set; Step S7, based on the best parameter geographic detector, study the factors of remote sensing ecological index evolution.

2. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 1, characterized in that, In step S2, the greenness index is represented by the normalized difference vegetation index NDVI NDVI with the following formula: ； Humidity index Wet Using the Gabor transform, the formula is as follows: ； wherein, , , , , , and are the surface reflectance of the wavebands in the multispectral image, respectively, , , , , , and are the corresponding coefficients, respectively. dryness index NDBSI from the bare soil index BI and the construction index IBI is synthesized, according to the following formula: ； ； ； wherein, and the surface reflectance of the respective multispectral image in the waveband. The temperature index is represented by the surface temperature LST and is given by the formula: ； wherein is luminance, is a wavelength in the thermal infrared band, is a physical constant, ; is the specific emissivity.

3. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 1, characterized in that, In step S3, the time series dataset is reconstructed, and the specific steps are as follows: Using time series harmonic method HANTS processing time series images of individual ecological factors from by fast fourier transform FFT decomposing the time profile into reconstructing the time series from the individual harmonic components, the formula being as follows: ； ； where, , and represent the original sequence, the reconstructed sequence and the error sequence, respectively, is the observation time of the remote sensing image, is the number of periods in the time series, , denotes the trigonometric component coefficients at the frequency , is the mean value of the entire sequence.

4. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 1, characterized in that, In step S4, based on the reconstructed time series dataset, dimensionless processing is carried out, and global normalization is carried out, the formula is as follows: ； wherein, is a specific ecological index is the original data of a pixel in time t is a dimensionless result, is a specific ecological index is the original data of a pixel in time t is the original data of a pixel in time min and max are minimum and maximum functions, respectively; The average value of multiple time phases is used as the ecological environment quality state of the ecological index in a certain period, the formula is as follows: ； wherein, is an ecological environment quality state for a specific period, , is an observation time of the remote sensing image; , is a trigonometric function component coefficient at a frequency of . according to PCA Principal component analysis and based on the components NDVI Eigenvector value adjustment of the first principal component PC1 direction as RSEI The value is used to assess the quality of the ecological environment; based on the first principal component. NDVI The direction of the eigenvector values ​​defines the direction of the first principal component, i.e., the eigenvector direction. NDVI If it is the positive direction PC1 If it's the positive direction, then it's the negative direction; otherwise, it's the opposite direction. The formula is as follows: ； If RSEI The direction is opposite to NDVI The direction is opposite to ； wherein is a direction function of the first principal component, is a direction of the first principal component.

5. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 1, characterized in that, In step S5, the remote sensing ecological index RSEI Model accuracy verification specifically involves: using the original... RSEI and after reconstruction RSEI The land cover types are categorized into five levels: excellent, good, average, poor, and very poor. RSEI Statistics on the percentage of different grades, and then the original RSEI With reconstruction RSEI By comparing the spatiotemporal distribution, a reconstruction can be performed. RSEI Accuracy verification.

6. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 5, characterized in that, The quality control QA band divides the pixel quality into four categories: high quality, quality to be considered, no pixel generated due to cloud, and no pixel generated due to other factors.

7. The long-time high-frequency ecological environment quality spatio-temporal differentiation and driving analysis method according to claim 1, characterized in that, In step S6, the Theil-Sen Median slope estimation formula is as follows: ； wherein is the median function; and is the median function; and Sen The significance of the slope estimate is determined by the Mann-Kendall trend test, which is formulated as follows: ； ； ； ； wherein, m is the number of time series, , is the time , corresponding data; is the sign function; Z is the trend test result, S is the statistic of the trend test; Hurst The exponential space distribution is divided into three cases: 0.5 < H < 1, which means that the time series has persistence, and the future trend is consistent with the past; H = 0.5 means that the sequence is a random sequence; 0 < H < 0.5 means that the sequence has anti-persistence, and the future trend is opposite to the past; The spatial Moran index is used to evaluate the correlation between a certain pixel and the surrounding area, the formula is as follows: ； where I is the Moran's index, with a value ranging from -1 to 1, and n is the total number of pixels, is a spatial weight matrix, and are the spatial attribute values of the i and j pixels, respectively, is the average. 8.The method of claim 1, wherein the method is characterized by, In step S7, the best parameter geographic detector can identify the best combination of spatial data discretization classification method and spatial scale parameter, accurately reveal the correlation between dependent variable and independent variable, mainly including factor detector module and interaction detector module; The factor detector module can detect the spatial differentiation of ecological quality and the explanatory power of the influencing factors, the formula is as follows: ； wherein, q L is the division level of the driving factor, is the number of samples of the hth level of the driving factor, is the total number of samples in the region; and is the variance of the hth level of the dependent variable and the total variance of the region, respectively, SSW is the sum of the intra-layer variances, SST is the total variance of the region. The interaction detector module is used to identify the interaction effect between different driving factors, and evaluate the explanatory power of the dependent variable Y, and judge the interaction type between the two driving factors, the specific steps are as follows: Step S71, calculating the independent variables respectively and Independent explanatory power for the dependent variable space differentiation and ; Step S72, calculate the explanatory power when the two factors act together ; Step S73, determining the type of interaction by comparing , and the relationships between the three.

Images