A method and system for environmental monitoring and phenological correlation processing based on GEE

Abstract

The present invention discloses a method and system for environmental monitoring and phenological correlation processing based on GEE. By obtaining environmental factors of the area to be studied, using Landsat 8 remote sensing data on the GEE platform for environmental monitoring according to the environmental factors to obtain spatio-temporal maps of the environmental factors, determining the temporal soil-adjusted vegetation phenology index from the maximum vegetation index and the minimum vegetation index, taking the temporal soil-adjusted vegetation phenology index as the phenological measurement value for phenological correlation analysis, analyzing the phenological correlation between the phenological measurement value and the index value corresponding to the surface environmental variables based on the support vector regression algorithm, using remote sensing data in GEE to monitor environmental changes and correlate them with phenological data, it is possible to identify the peaks and valleys of each parameter in the environmental factors, revealing a significant correlation between TSPI and environmental factors, improving the efficiency of diverse analysis and processing of environmental monitoring and phenological correlation, and enhancing the accuracy of phenological correlation assessment.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing applications, and particularly relates to a method and system for environmental monitoring and phenological correlation processing based on GEE. Background Art

[0002] Over time and space, environmental changes have had a significant impact on agriculture, forestry, water resource management, public health, and ecosystems. Understanding and monitoring these fluctuations is crucial for making informed decisions. Although ground-based environmental monitoring methods provide valuable information about soil moisture, temperature, water resources, air quality, and land cover trends, their scope and coverage are often limited. Using satellite-based technologies at broad spatial and temporal scales is essential for ecological research. Environmental monitoring is crucial for protecting the planet, maintaining ecological balance, and improving the quality of human life. Traditionally, it has relied on ground and meteorological data. For example, the Netatmo weather station in Berlin is used to collect data on temperature, humidity, air pressure, carbon dioxide levels, and noise, which are transmitted via Wi-Fi every five minutes. By juxtaposing this information with reference data, the importance of ground stations in urban environmental research is emphasized. Similarly, raster format maps of crop-specific land cover are generated using field data. Researchers have developed an advanced Internet of Things system for comprehensive weather monitoring, providing real-time access to various environmental data, including temperature, humidity, wind speed, humidity, light intensity, ultraviolet radiation, and carbon monoxide levels. This system uses multiple ground sensors to collect data and presents it on a dedicated web page through charts and statistics. This setup enhances accessibility and makes monitoring and reference simple and clear. Although these data can provide valuable insights into soil moisture, temperature, or land cover trends, they are limited to specific areas and require significant time, money, and labor inputs. This emphasizes the need for remote sensing satellite data as an effective alternative. Remote sensing sensors are capable of quickly collecting comprehensive spectral data across regional and global scales, providing extensive and dynamic observation capabilities.

[0003] Efficient environmental management relies on accessing a comprehensive database within a specified time. By using remote sensing technologies and methods, various principles for capturing the electromagnetic characteristics of the Earth's surface can be utilized to extract key indicators. Through indices derived from satellite data, this expertise provides a valuable resource for analyzing ecological fluctuations in different fields. Google Earth Engine (GEE) is a valuable cloud-based remote sensing platform that offers rich data resources and powerful capabilities for conducting change detection studies. This platform has played an important role in numerous studies, providing diverse analysis and processing capabilities. For example, using GEE to monitor changes in the Normalized Difference Vegetation Index (NDVI) for crop mapping reveals growth patterns; similarly, using GEE and Landsat 8 images to establish a positive correlation between Land Surface Temperature (LST) and NDVI helps to assess land degradation and land cover classification; forest degradation monitoring is carried out through time series analysis of GEE, effectively mapping with minimal uncertainty to support conservation efforts; the relationship between LST and forest changes within the Paphos Forest highlights the utility of NDVI and LST in identifying potential forest decline using GEE; using an integrated model and correlation analysis with Landsat-8 data on GEE to predict LST, this innovative approach enhances the understanding of the impacts of climate change. However, using GEE to study key factors such as vegetation cover, land surface temperature, and soil moisture within different time frames provides valuable insights into environmental changes. These insights are crucial for studying various fields, including ecological events such as floods, public health, and crisis management. Therefore, it is essential to adopt an interdisciplinary approach that transcends traditional scientific boundaries. In this context, specific overarching concepts, such as phenology, are relevant across multiple scientific disciplines. Phenology is a bridge between environmental observation and remote sensing, involving the monitoring of periodic biological events and the assessment of ecosystem responses to climate change. In addition to measurements, this framework provides a valuable perspective on ecological dynamics through satellite data. Scholars have emphasized a strong correlation between phenology and environmental factors, highlighting its significance in understanding ecosystem processes.

[0004] Therefore, there is an urgent need to provide a method and system for environmental monitoring and phenology correlation processing based on GEE to solve the above-mentioned existing technical problems. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for environmental monitoring and phenology correlation processing based on GEE that improves the efficiency and rationality of environmental monitoring and phenology correlation processing. Landsat 8 data within the GEE platform is used to monitor environmental factors and calculate phenology indices, enhancing the usefulness of phenology indices in understanding seasonal changes. The following specific solutions are specifically adopted to achieve this.

[0006] In a first aspect, the present invention provides a method for environmental monitoring and phenological correlation processing based on GEE, comprising the following steps:

[0007] Obtain environmental factors of the area to be studied, and use Landsat 8 remote sensing data on the GEE platform for environmental monitoring based on the environmental factors to obtain spatio-temporal maps of the environmental factors, wherein the environmental factors include meteorological data and surface environmental variables, the surface environmental variables include soil-adjusted vegetation index, surface temperature, and soil moisture index, and the spatio-temporal maps include time series maps and timeline maps respectively corresponding to the surface environmental variables;

[0008] Determine the time soil-adjusted vegetation phenological index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal maps, and use the time soil-adjusted vegetation phenological index as the phenological measurement value for phenological correlation analysis;

[0009] Analyze the phenological correlation between the phenological measurement value and the index values corresponding to the surface environmental variables based on the support vector regression algorithm, wherein the index values include minimum value, maximum value, average value, and difference.

[0010] As a preference of the above technical solution, extracting the surface temperature using Landsat 8 includes:

[0011] Convert the digital DN to the top-of-atmosphere TOA radiance value, and convert the top-of-atmosphere TOA radiance value to the brightness temperature BT. Among them, the expression for obtaining the conversion of the digital DN to the top-of-atmosphere TOA radiance value is formula (1), and the expression for converting the top-of-atmosphere TOA radiance value to the brightness temperature BT is formula (2):

[0012] (1)

[0013] (2)

[0014] Among them, represents the radiance value of band 10, represents the increased brightness value of radiance band 10, DN represents the quantized standard product pixel value, represents the correction value of band 10 and takes the value of 0.29, , respectively represent specific constants of band 10. To convert the surface temperature LST from Kelvin to Celsius, 273.15 needs to be subtracted;

[0015] Obtain the vegetation index NDVI and the vegetation ratio , and calculate the surface emissivity according to the vegetation index NDVI and the vegetation ratio , and the corresponding expression is formula (5):

[0016] (3)

[0017] (4)

[0018] (5)

[0019] Wherein, represents the land surface emissivity, represents the vegetation fraction applied to the formula in the raster calculator, and 0.986 is the correction value of the formula, represents band 4, represents band 5;

[0020] Estimate the land surface temperature LST based on the top-of-atmosphere TOA radiance value and the surface emissivity. Wherein, determine the land surface temperature LST according to Planck's law, and derive formula (6) according to Planck's law:

[0021] (6)

[0022] (7)

[0023] Wherein, represents the data of band 10 in Landsat 8, represents the wavelength of the emitted radiation, corresponds to Planck's constant and has a value of , represents Boltzmann's constant and has a value of , and c represents the light beam and has a value of .

[0024] As a preference of the above technical solution, calculate the soil-adjusted vegetation index SAVI according to the vegetation index NDVI to evaluate the presence of green vegetation. The expression for calculating the soil-adjusted vegetation index SAVI in Landsat 8 is formula (8):

[0025] (8)

[0026] Wherein, the assignment of the L value represents the change level of green vegetation. In areas without any green cover, the L value is set to 1; in areas with moderate greening, the L value is 0.5; in areas with dense vegetation, the L value drops to 0. The L value is used to reflect the vegetation index NDVI, and the NDVI value ranges from -1.0 to 1.0.

[0027] As a preference of the above technical solution, the soil moisture index is evaluated and calculated using formula (9):

[0028] (9)

[0029] Among them, and represent the maximum and minimum land surface temperatures of each image, respectively.

[0030] As an optimization of the above technical solution, the Time Soil Adjusted Phenological Index (TSPI) is used to quantify the difference between the maximum and minimum vegetation activities during the phenological cycle. The calculation formula of TSPI is:

[0031] (10)

[0032] Among them, and represent the maximum and minimum values of SAVI each year, respectively. TSPI represents the temporal variation of land cover, 0 represents no change, and 1 represents the maximum change.

[0033] As an optimization of the above technical solution, the phenological correlation between the phenological measurement values and the index values corresponding to the surface environmental variables is analyzed based on the Support Vector Regression (SVR) algorithm, including:

[0034] The Support Vector Regression (SVR) algorithm uses a kernel function to map data to a high dimension to find a hyperplane that best separates the data points. Among them, SVR minimizes the error through a regularization term and slack variables, achieving a balance between maximizing the margin and minimizing the error. The minimized loss function is:

[0035] (11)

[0036] Among them, represents the feature importance, represents the offset decision boundary, and provide the flexibility of data fitting, is used to balance error minimization and overfitting prevention, defines the allowable deviation from the true value, and n represents the training examples;

[0037] The R-squared value in SVR is used to measure the degree to which the independent variable explains the variance of the dependent variable and is a key indicator for evaluating the goodness of fit of the model in regression analysis. The corresponding formula is:

[0038] (12)

[0039] Among them, is the sum of squares of the errors between the observed values and the predicted values, is the sum of squares and is used to measure the total variance of the dependent variable.

[0040] Preferably, as the above technical solution, the support vector regression algorithm is used to analyze and compare the meteorological data with the output environmental factors to verify the analysis results corresponding to the phenological correlation. Among them, the meteorological data includes the monthly average values of temperature, humidity, and precipitation in the area to be studied.

[0041] Preferably, as the above technical solution, the environmental factors of the area to be studied are obtained, including:

[0042] The remote sensing image corresponding to the environmental factors is preprocessed. The preprocessing includes masking clouds and shadows in the remote sensing image. Among them, the functions cloudMaskL457 and maskL8sr are used to mask low-quality observations, and a bit mask is created using the pixel quality assessment band. Shadows and clouds are identified through bit operations.

[0043] In a second aspect, the present invention also provides an environmental monitoring and phenological correlation processing system based on GEE, which is applied to the above-mentioned environmental monitoring and phenological correlation processing method based on GEE, including:

[0044] An environmental factor acquisition module, which is used to acquire the environmental factors of the area to be studied, and use Landsat 8 remote sensing data on the GEE platform to conduct environmental monitoring based on the environmental factors to obtain the spatio-temporal maps of the environmental factors. Among them, the environmental factors include meteorological data and surface environmental variables. The surface environmental variables include soil-adjusted vegetation index, surface temperature, and soil moisture index. The spatio-temporal maps include the time series maps and timeline maps corresponding to the surface environmental variables respectively;

[0045] A phenological index determination module, which is used to determine the time soil-adjusted vegetation phenological index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal maps, and use the time soil-adjusted vegetation phenological index as the phenological measurement value for phenological correlation analysis;

[0046] A phenological correlation analysis module, which is used to analyze the phenological correlation between the phenological measurement value and the index values corresponding to the surface environmental variables based on the support vector regression algorithm. Among them, the index values include minimum value, maximum value, average value, and difference.

[0047] The present invention provides a method and system for environmental monitoring and phenological correlation processing based on GEE. By obtaining environmental factors of the area to be studied, and using Landsat 8 remote sensing data on the GEE platform according to the environmental factors for environmental monitoring to obtain the spatio-temporal map of the environmental factors, determining the temporal soil-adjusted vegetation phenology index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal map, and using the temporal soil-adjusted vegetation phenology index as the phenological measurement value for phenological correlation analysis, analyzing the phenological correlation between the phenological measurement value and the index value corresponding to the surface environmental variable based on the support vector regression algorithm, using remote sensing data in GEE to monitor environmental changes and correlate them with phenological data, the peaks and valleys of each parameter in the environmental factors can be identified. SVR analysis reveals a significant correlation between TSPI and environmental factors, improving the efficiency of diverse analysis and processing of environmental monitoring and phenological correlation, and enhancing the accuracy of phenological correlation assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and thus should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 It is a flowchart of the method for environmental monitoring and phenological correlation processing based on GEE provided by the present invention;

[0050] Figure 2 It is a monthly average map of precipitation, air humidity and temperature in YLYLM Province from 2014 to 2021 provided by the present invention;

[0051] Figure 3 It is a TSPI map of YLYLM Province from 2014 to 2021 provided by the present invention;

[0052] Figure 4 It is an average TSPI timeline map of YLYLM Province (2014 - 2021) provided by the present invention;

[0053] Figure 5 It is a flowchart of the system for environmental monitoring and phenological correlation processing based on GEE provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

[0055] Referring to Figure 1 , the present invention provides a method for environmental monitoring and phenological correlation processing based on GEE, comprising the following steps:

[0056] S10: Obtain environmental factors of the area to be studied, and use Landsat8 remote sensing data on the GEE platform for environmental monitoring according to the environmental factors to obtain spatio-temporal maps of the environmental factors, wherein the environmental factors include meteorological data and surface environmental variables, the surface environmental variables include soil-adjusted vegetation index, land surface temperature and soil moisture index, and the spatio-temporal maps include time series maps and timeline maps respectively corresponding to the surface environmental variables;

[0057] S11: Determine the time soil-adjusted vegetation phenological index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal maps, and use the time soil-adjusted vegetation phenological index as the phenological measurement value for phenological correlation analysis;

[0058] S12: Analyze the phenological correlation between the phenological measurement value and the index values corresponding to the surface environmental variables based on the support vector regression algorithm, wherein the index values include minimum value, maximum value, average value and difference.

[0059] In this embodiment, remote sensing data and Google Earth Engine were utilized to analyze the environmental fluctuations in YLYLM Province from 2014 to 2021. We focused on monitoring environmental variables and their correlations with phenology, adopting the Support Vector Regression (SVR) technique. Landsat 8 satellite data was used to generate time series maps and timelines of land cover, temperature, and soil moisture, using the Soil-Adjusted Vegetation Index (SAVI), Land Surface Temperature (LST) anomaly, and Soil Moisture Index (SMI). Subsequently, the Temporal Soil-Adjusted Phenology Index (TSPI) was calculated to monitor annual vegetation changes, and the correlation with specified parameters was analyzed using the R-squared value in the SVR analysis. This invention reveals significant changes in environmental attributes and a strong correlation with the TSPI index. Soil moisture peaks in late winter and spring but decreases in summer, reaching its maximum in 2018. Vegetation is most abundant in mid-spring and least in winter, with significant greening in 2019. Summer temperatures are the highest and winter temperatures are the lowest, with the smallest interannual variations. Spatial pattern monitoring shows that surface temperature continuously increases from the north and east to the south and west, correlated with the decreasing trends in vegetation and soil moisture levels. Regression analysis indicates a robust association between TSPI and environmental variables, with an R-squared value of 0.84 for LST, 0.91 for SAVI, and 0.79 for SMI. These highlight the effectiveness of remote sensing methods, such as time series satellite imagery and simplified indices, for large-scale ecological analysis using the GEE platform, and emphasize the potential of TSPI as a key indicator for future environmental management research.

[0060] It should be noted that the ecosystem has rich land cover diversity and is affected by spatial differences and seasonal dynamics. The area to be studied is YLYLM Province, which is an ideal research object due to its diverse climate and landscape. The coordinate range of YLYLM Province is from 33°19' north latitude to 35°54' north latitude and from 38°45' east longitude to 30°52' east longitude, with an area of 20,133 square kilometers and an average altitude of 1,218 meters. The northern and eastern regions of the province are mountainous areas, mainly semi-forested areas with a cold and semi-humid climate, while the western and southern regions are arid deserts with a hot and dry climate. Continuous Landsat 8 remote sensing data was used. The reason for choosing these data is that since 2013, it has proven its reliability, accessibility, and suitable spectral and spatial resolution through GEE. These data can continuously monitor environmental variables through relevant spectral indices, thus contributing to a detailed analysis of surface attributes and trends. Detailed resolution information is shown in Table 1 in the appendix.

[0061] Table 1: Resolution of Landsat 8

[0062]

[0063] Meteorological data is incorporated into this embodiment and compared with the output environmental factors through regression analysis to verify the results. The dataset includes the monthly averages of air temperature, humidity, and precipitation obtained from weather stations acquired from the meteorological bureau. The environmental factors of the area to be studied are obtained, including: performing data preprocessing on the remote sensing images corresponding to the environmental factors, where the preprocessing includes masking clouds and shadows in the remote sensing images. Among them, the functions cloudMaskL457 and maskL8sr are used to mask low-quality observations, and a bit mask is created using the pixel quality assessment band, and shadows and clouds are identified through bit operations. Using the SAVI index, LST anomaly, and SMI index, time series plots and timeline plots of three key factors: land cover, temperature, and soil moisture are extracted using Landsat 8 data on the GEE platform. These maps illustrate the annual trends and spatial patterns of the entire study area by showing the distribution of environmental variables. In addition, the corresponding charts show the temporal fluctuations, highlighting the peaks and troughs of each parameter over eight consecutive years from 2014 to 2021. In addition, the correlations between these datasets and meteorological data are also analyzed to confirm the obtained maps and charts.

[0064] In the subsequent phenological correlation processing, the TSPI is analyzed, which is a phenological measurement derived from the SAVI-max maximum vegetation index and the SAVI-min minimum vegetation index. Its correlations with each index value (minimum, maximum, average, and difference) are analyzed using support vector regression SVR.

[0065] It should be understood that by obtaining the environmental factors of the area to be studied, obtaining the spatio-temporal maps of the environmental factors through environmental monitoring using Landsat 8 remote sensing data on the GEE platform according to the environmental factors, determining the time soil-adjusted vegetation phenology index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal maps, using the time soil-adjusted vegetation phenology index as the phenological measurement for phenological correlation analysis, analyzing the phenological correlations between the phenological measurement and the index values corresponding to the surface environmental variables based on the support vector regression algorithm, using the remote sensing data in GEE to monitor environmental changes and correlate them with phenological data, the peaks and troughs of each parameter in the environmental factors can be identified. The SVR analysis reveals a significant correlation between the TSPI and the environmental factors, improving the efficiency of diverse analysis and processing of environmental monitoring and phenological correlations, and enhancing the accuracy of phenological correlation assessment.

[0066] Optionally, extracting the land surface temperature using Landsat 8 includes:

[0067] Convert the digital number DN to the top-of-atmosphere (TOA) radiance value, and convert the top-of-atmosphere (TOA) radiance value to the brightness temperature (BT). Among them, the expression for obtaining the conversion of the digital number DN to the top-of-atmosphere (TOA) radiance value is formula (1), and the expression for converting the top-of-atmosphere (TOA) radiance value to the brightness temperature (BT) is formula (2):

[0068] (1)

[0069] (2)

[0070] Among them, represents the radiance value of band 10, represents the increased brightness value of radiance band 10, DN represents the quantized standard product pixel value, represents the correction value of band 10 and takes the value of 0.29, 、 respectively represent the specific constants of band 10. To convert the land surface temperature (LST) from Kelvin to Celsius, 273.15 needs to be subtracted;

[0071] Obtain the normalized difference vegetation index (NDVI) and vegetation fraction According to the normalized difference vegetation index (NDVI) and vegetation fraction Calculate the land surface emissivity, and the corresponding expression is formula (5):

[0072] (3)

[0073] (4)

[0074] (5)

[0075] Among them, represents the land surface emissivity, represents the vegetation fraction applied to the formula in the raster calculator, and 0.986 is the correction value of the formula, represents band 4, represents band 5;

[0076] Estimate the land surface temperature (LST) based on the top-of-atmosphere (TOA) radiance value and the land surface emissivity. Among them, determine the land surface temperature (LST) according to Planck's law, and derive formula (6) according to Planck's law:

[0077] (6)

[0078] (7)

[0079] Among them, Represents the data of band 10 in Landsat 8, represents the wavelength of the emitted radiation, corresponds to the Planck constant and has a value of , represents the Boltzmann constant and has a value of , c represents the light beam and has a value of .

[0080] In this embodiment, LST refers to the thermal energy released from the Earth's surface and captured in the atmosphere. Landsat 8 OLI / TIRS has specific bands to extract such thermal data anomalies. Top of Atmosphere (TOA) reflectance: The TOA reflectance is the reflectance of a satellite remote sensing image when it reaches the ground after passing through the atmosphere, without considering the effects of light absorption, scattering, and reflection by the surface and the atmosphere. Therefore, the TOA reflectance is the most basic data obtained from satellite remote sensing images.

[0081] Optionally, the Soil-Adjusted Vegetation Index (SAVI) is calculated based on the Normalized Difference Vegetation Index (NDVI) to evaluate the presence of green vegetation. The expression for calculating SAVI in Landsat 8 is formula (8):

[0082] (8)

[0083] where the assignment of the L value represents the change level of green vegetation. In areas without any green cover, the L value is set to 1; in areas with moderate greening, the L value is 0.5; in areas with dense vegetation, the L value drops to 0. The L value is used to reflect the vegetation index NDVI, and the NDVI value ranges from -1.0 to 1.0.

[0084] In this embodiment, the soil moisture index is evaluated and calculated using formula (9):

[0085] (9)

[0086] where and represent the maximum and minimum surface temperatures of each image respectively. The evaluation of SMI has attracted extensive attention in previous studies involving remote sensing parameters. A reliable method applied in this case is to utilize LST.

[0087] Optionally, the Time-Series Soil-Adjusted Vegetation Phenology Index (TSPI) is used to quantify the difference between the maximum and minimum vegetation activities during the phenological cycle. The calculation expression of TSPI is:

[0088] (10)

[0089] where and represent the maximum and minimum values of SAVI per year respectively, TSPI represents the temporal change of land cover, 0 represents no change, and 1 represents the largest change. The phenological index is a valuable tool for monitoring the development that occurs annually or seasonally throughout the growth cycle. It helps to track the response of ecosystems to environmental fluctuations. Using the temporal SAVI phenological index (TSPI), which quantifies the difference between the maximum and minimum vegetation activities in the phenological cycle, this temporal index provides a concise insight into the phenological changes caused by changes in environmental factors.

[0090] Optionally, analyze the phenological correlation between the phenological measurement values and the index values corresponding to the surface environmental variables based on the support vector regression algorithm, including:

[0091] The support vector regression algorithm SVR uses a kernel function to map the data to a high dimension to find a hyperplane that best separates the data points; among them, SVR minimizes the error through a regularization term and slack variables, achieving a balance between maximizing the margin and minimizing the error. The minimized loss function is:

[0092] (11)

[0093] where represents the feature importance, represents the offset decision boundary, and provides the flexibility of data fitting, is used to balance error minimization and overfitting prevention, defines the allowable deviation from the true value, and n represents the training examples;

[0094] The R-squared value in SVR is used to measure the degree to which the independent variable explains the variance of the dependent variable and is a key indicator for evaluating the goodness of fit of the model in regression analysis. The corresponding expression is:

[0095] (12)

[0096] where is the sum of the squares of the errors between the observed values and the predicted values, is the sum of squares and is used to measure the total variance of the dependent variable.

[0097] In this embodiment, the support vector regression algorithm is used to analyze and compare meteorological data with the output environmental factors to verify the analysis results corresponding to the phenological correlation. Among them, the meteorological data includes the monthly averages of temperature, humidity, and precipitation in the area to be studied. Statistical analysis used support vector regression (SVR), which effectively investigated the relationships between the variables encountered. SVR uses kernel functions to map data into higher dimensions, aiming to find a hyperplane that optimally separates data points. It minimizes the error through regularization terms and slack variables, achieving a balance between maximizing the margin and minimizing the error.

[0098] Specifically, environmental factor monitoring includes multi-temporal LST based on Landsat 8 data, multi-temporal SAVI based on Landsat 8 data, and multi-temporal SMI based on Landsat 8 data. Maps and timelines of three key factors (LST, SAVI, and SMI) during the entire measurement period provided insights into their fluctuations and evolution. The correlation study results between these factors and TSPI were demonstrated through regression analysis, aiming to clarify the relationship between the environment and phenology. This analysis method enables us to quantify and interpret the correlations between these variables, revealing potential ecological patterns and dynamics.

[0099] Among them, multi-temporal imagery refers to the characteristics of a set of remote sensing images in a time series, including image data acquired at different times such as satellite remote sensing. These images can be used for dynamic analysis and change detection. LST reaches its maximum values in June, July, and August, while reaching its lowest levels in December, January, and February. The LST chart shows that the fluctuations are minimal across different years, and the annual charts always depict very similar and invariant patterns. At the provincial level, the LST distribution shows a trend of increasing from the mountainous areas in the north and east to the desert areas in the south and west. The 2021 LST map shows a significant increase in the high-temperature areas expanding westward, while the 2015 map clearly shows a significant decrease. As shown in Table 2, the highest LST recorded in 2021 is in sharp contrast to the lowest point observed in 2017, with the lowest and highest average levels in 2015 and 2021 respectively.

[0100] Table 2: Maximum, minimum, average, and difference of LST (°C) in YLYLM Province from 2014 to 2021

[0101]

[0102] The multi-temporal SAVI based on Landsat 8 data includes: the SAVI timeline for the entire measurement period, the annual average SAVI map, and the significant changes in green land cover over the years. These maps show a decreasing trend in the greening area from the highlands in the northern and eastern regions of the province to the arid and desert areas in the southern and western regions. As shown in Table 3, the monthly fluctuations continuously exhibit higher SAVI values in spring, while the lowest values are recorded in winter. The annual maps show changes, with a significant increase in the greening area in 2019 and 2020 compared to other years. The average green land level was the lowest in 2014. In addition, the annual change was the largest in 2015 and the smallest in 2021.

[0103] Table 3: Maximum, minimum, average, and difference of SAVI index values in YLYLM Province from 2014 to 2021

[0104]

[0105] The multi-temporal SMI based on Landsat 8 data includes: The soil moisture maps show that the SMI levels are the highest in November and December, while the values are the lowest in July and August, and a comprehensive breakdown of the annual SMI values is presented in Table 4. The SMI maps show a gradual upward trend from the southern and western regions to the central, northern, and eastern regions. Notably, there is a strong correlation between this index and the vegetation level within 5 months. The peaks and valleys in the SAVI map are closely aligned with the highest and lowest values in the SMI map.

[0106] Table 4: Maximum, minimum, average, and difference of SMI index in YLYLM Province from 2014 to 2021

[0107]

[0108] Soil-adjusted vegetation phenology index (TSPI): TSPI is calculated by determining the maximum and minimum values of the SAVI index each year. Figure 2 Describes the distribution pattern of TSPI, while the subsequent Table 5 and Figure 3 shows the average value for each year in the study area. In 2014 and 2021, the TSPI values both reached extremes, marking the peaks and valleys of the greening fluctuations. As shown in the map, the distribution of this index has no obvious pattern. The Figure 4 depicts a continuous and gradual downward trend in the annual average amount, except in 2017 when the index value showed a significant increase.

[0109] Table 5: Average TSPI in YLYLM Province from 2014 to 2021

[0110]

[0111] Specifically, the statistical analysis includes environmental parameters and phenological indices, environmental parameters and ground meteorological data. The environmental parameters and phenological indices include LST and TSPI, SAVI and TSPI, SMI and TSPI. The correlation between the mean value of the specified phenological index and the three discussed environmental variables (LST, SAVI, and SMI) was examined through SVR analysis. Nonlinear regression was performed using the RBF kernel, with "C = 100" set as regularization, "gamma = 0.1" set as the influence range, and "epsilon =.1" set as the prediction tolerance. The SVR analysis of different LST metrics (minimum, maximum, difference, and mean) and the annual average TSPI value showed a significant correlation. The R-squared values between TSPI and LST were 0.84 at minimum, 0.76 at maximum, 0.79 for the difference, and 0.67 for the mean, as shown in Table 6 for details.

[0112] Table 6: Results of the regression analysis between LST values and TSPI

[0113]

[0114] Table 7 shows the results of the SVR analysis of the mean values of TSPI and SAVI each year, highlighting the strong correlation between these parameters. This alignment strengthens the relationship between phenology and the greening index. The R-squared values for the minimum, maximum, difference, and mean SAVI of TSPI were 0.91, 0.62, 0.64, and 0.74 respectively. These values quantify the proportion of the variability of TSPI that is explained by each corresponding aspect of SAVI.

[0115] Table 7: Results of the correlation analysis between SAVI values and TSPI

[0116]

[0117] The regression analysis shows a moderate correlation between the SMI value and TSPI, with an R-squared value of approximately 0.4, as shown in Table 8 for details. In addition, the peak on the SMI map strongly corresponds to the peak on the SAVI map within a specific time range. This indicates a significant correlation between SMI and SAVI, with a time lag of approximately five months due to the greening period.

[0118] Table 8: Results of the correlation analysis between SMI values and TSPI

[0119]

[0120] Refer to Figure 5 , the present invention also provides an environmental monitoring and phenological correlation processing system based on GEE, which is applied to the above-mentioned environmental monitoring and phenological correlation processing method based on GEE, and includes:

[0121] An environmental factor acquisition module is used to acquire the environmental factors of the area to be studied, and based on the environmental factors, environmental monitoring is carried out using Landsat 8 remote sensing data on the GEE platform to obtain the spatio-temporal maps of the environmental factors. Among them, the environmental factors include meteorological data and surface environmental variables. The surface environmental variables include soil adjusted vegetation index, land surface temperature, and soil moisture index. The spatio-temporal maps include the time series maps and timeline maps respectively corresponding to the surface environmental variables;

[0122] A phenological index determination module is used to determine the time soil adjusted vegetation phenological index from the maximum vegetation index and the minimum vegetation index according to the spatio-temporal maps, and use the time soil adjusted vegetation phenological index as the phenological measurement value for phenological correlation analysis;

[0123] A phenological correlation analysis module is used to analyze the phenological correlation between the phenological measurement value and the index values corresponding to the surface environmental variables based on the support vector regression algorithm. Among them, the index values include minimum value, maximum value, average value, and difference.

[0124] In this embodiment, the SVR analysis was used to examine the correlation between the average values of the generated parameters (LST, SAVI, SMI) and the meteorological data (monthly average values of air temperature, humidity, and precipitation). There is a strong relationship between ground measurements and remote sensing data. Specifically, the R-squared value indicates a substantial correlation; the correlation between LST and air temperature is 0.89, the correlation between SAVI and air humidity is 0.78, and the correlation between SMI and precipitation is 0.72, which confirms the reliability of the generated factors. The spatio-temporal maps of environmental factors were generated using advanced remote sensing technology, effectively illustrating the dynamics and trends of variables such as temperature, vegetation, and soil moisture in YLYLM Province. The spatial analysis using LST, SAVI, and SMI maps shows that from the mountainous areas in the northeast to the desert areas in the southwest, the land surface temperature shows an upward trend, while the green areas and soil moisture show a downward trend. The analysis of the greening time depicted by the SAVI timeline map shows that the peak in spring is related to the optimal air temperature and rainfall, while the minimum value in winter is consistent with previous studies. According to the observations, the land surface temperature is continuously the highest in summer, the lowest in winter, and has the smallest interannual fluctuations. The assessment of soil moisture through SMI shows that the maximum values at the end of winter and in spring are in sharp contrast to the minimum values in summer, and the ground data confirm these results.

[0125] It should be noted that the average values of LST, SAVI, and SMI in 2021, 2019, and 2018 were the highest, while those in 2015, 2014, and 2020 were the lowest. The TSPI index was used to evaluate the temporal vegetation phenology, and the TSPI index utilized the annual minimum and maximum SAVI values as reliable indicators. Due to the semi-arid conditions of the study area, SAVI was selected for this index as it effectively addressed the soil brightness issue by incorporating the L value into the NDVI formula. TSPI provided an accurate method to quantify the impact of environmental factors on the phenological cycle, and the corresponding timeline graph showed a generally decreasing trend except in 2017. The greening change was most significant in 2014 and reached the lowest point in 2021. Subsequently, SVR analysis was conducted to explore the correlation between environmental factors (including maximum, minimum, average, and differential values) and the annual average value of TSPI, revealing a significant correlation between TSPI and these variables within the specified study area. The SVR R-squared value for LST was 0.84, for SAVI was 0.91, and for the average SMI was 0.79.

[0126] There was a significant correlation between SMI and greenness within a five-month time frame. The peak soil moisture in late November 2018 coincided with the maximum greenness observed in April 2019. Conversely, the decrease in soil water content in 2020 led to a reduction in greening in the SAVI map in 2021. This correlation was due to favorable temperature conditions promoting plant growth and being facilitated by optimal soil moisture. The time lag was a result of the green cycle and was considered when estimating the correlation between the SMI index and TSPI. These results indicate that sufficient rainfall and high soil moisture in autumn and winter will result in lush greenery in spring. Conversely, insufficient rainfall will lead to a reduction in greening. The results were verified through the correlation between the extracted variables and ground data, and the R-squared values showed significant correlations: 0.89 between LST and air temperature, 0.78 between SAVI and air humidity, and 0.72 between SMI and precipitation. The analysis results highlighted the effectiveness of remote sensing technology in accurate environmental monitoring. Additionally, the strong correlation between environmental factors and vegetation phenology indicates that the TSPI index has the potential to be a reliable environmental factor in future research, thus potentially reducing the need for large datasets in time series analysis.

[0127] It should be understood that remote sensing indices in GEE are used to monitor environmental changes and correlate them with phenological data, track the spatio-temporal changes in green land cover, temperature, and soil moisture, analyze the time series graphs and charts derived from SAVI, LST, and SMI data. Such visualization can track the changes in environmental conditions over time, identify the peaks and valleys of each parameter, and reveal the spatial trends across the province. Characterized by heterogeneous land cover and weather patterns, these environmental parameters exhibit different patterns. Notably, greenness and soil moisture show an upward trend from the southern and western regions towards the northern and eastern regions, while surface temperature shows a downward trend, indicating a significant negative correlation between them and temperature. Additionally, time series analysis reveals significant fluctuations in land cover and soil moisture over time, while the changes in surface temperature are relatively less obvious. SVR analysis reveals a significant correlation between TSPI and environmental factors, effectively tracking the changes in vegetation phenology. This provides valuable insights for future ecological research. TSPI can serve as a key environmental parameter in various algorithms, including those for risk assessment and environmental prediction.

[0128] In all the examples shown and described here, any specific values should be construed as merely exemplary and not as limitations. Thus, other examples of the exemplary embodiments may have different values.

[0129] It should be noted that like reference numerals and letters refer to like items in the following figures. Thus, once an item is defined in one figure, further definition and explanation thereof are not required in subsequent figures.

[0130] The above-described embodiments merely represent several implementation manners of the present invention. Their descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all fall within the protection scope of the present invention.

Claims

1. A method for environmental monitoring and phenological correlation processing based on GEE, characterized in that: The following steps are involved: Acquire environmental factors of the area to be studied, and use Landsat 8 remote sensing data to perform environmental monitoring on the GEE platform based on the environmental factors to obtain a spatiotemporal map of the environmental factors, wherein the environmental factors include meteorological data and surface environmental variables, the surface environmental variables include soil adjusted vegetation index, surface temperature and soil moisture index, and the spatiotemporal map includes a time series map and a timeline map corresponding to the surface environmental variables respectively; Determining a temporal soil-adjusted vegetation phenological index from a maximum vegetation index and a minimum vegetation index according to the spatiotemporal graph, and using the temporal soil-adjusted vegetation phenological index as a phenological measurement value for phenological correlation analysis; Analyzing the phenological correlation between the phenological measurement value and the index value corresponding to the surface environmental variable based on a support vector regression algorithm, wherein the index value includes a minimum value, a maximum value, an average value and a difference value; Using Landsat 8 to extract land surface temperature, including: The digital DN is converted into the TOA radiation value at the top of the atmosphere, and the TOA radiation value at the top of the atmosphere is converted into the brightness temperature BT. The expression for converting the digital DN into the TOA radiation value at the top of the atmosphere is formula (1), and the expression for converting the TOA radiation value at the top of the atmosphere into the brightness temperature BT is formula (2): （1） （2） in, Indicates the radiation value of band 10, represents the increased brightness value of radiation band 10, Represents the quantized standard product pixel value, It represents the correction value of band 10 and its value is 0.

29. , They represent the specific constants of band 10. To convert the land surface temperature LST from Kelvin to Celsius, you need to subtract 273.15; Get vegetation index NDVI and vegetation ratio , based on the vegetation index NDVI and vegetation ratio Calculate the surface emissivity, the corresponding expression is formula (5): （3） （4） （5） in, is the land surface emissivity, Indicates the vegetation ratio used in the formula in the raster calculator. 0.986 is the correction value of the formula. Indicates band 4, Indicates band 5; The surface temperature LST is estimated based on the TOA radiation value at the top of the atmosphere and the surface emissivity, wherein the surface temperature LST is determined according to Planck's law, and formula (6) is derived according to Planck's law: （6） （7） in, Represents the data of band 10 in Landsat 8. represents the wavelength of the emitted radiation, corresponds to Planck's constant and has a value of , represents the Boltzmann constant and its value is , Represents a beam and takes a value for; The soil-adjusted vegetation index (SAVI) is calculated based on the vegetation index NDVI to assess the presence of green vegetation. The expression for calculating the soil-adjusted vegetation index (SAVI) in Landsat 8 is formula (8): （8） The L value distribution indicates the level of change in green vegetation. In areas without any green cover, the L value is set to 1; in areas with moderate greening, the L value is 0.5; in areas with dense vegetation, the L value drops to 0. The L value is used to reflect the vegetation index NDVI, which ranges from -1.0 to 1.

0. The soil moisture index is evaluated and calculated using formula (9): （9） in, and Respectively represent the maximum and minimum surface temperatures of each image; The time-adjusted soil vegetation phenology index (TSPI) is used to quantify the difference between the maximum and minimum vegetation activities in the phenological cycle. The calculation expression of TSPI is: （10） in, and They represent the maximum and minimum values ​​of SAVI each year, respectively, and TSPI represents the temporal change of land cover, with 0 indicating no change and 1 indicating the maximum change; Analyzing the phenological correlation between the phenological measurement value and the index value corresponding to the surface environmental variable based on a support vector regression algorithm includes: The support vector regression algorithm SVR uses a kernel function to map the data to high dimensions to find a hyperplane that best separates the data points. SVR minimizes the error through regularization terms and slack variables, achieving a balance between maximizing the margin and minimizing the error. The minimized loss function is: （11） in, represents the feature importance, represents the offset decision boundary, and Provides flexibility in data fitting, Used to balance error minimization and overfitting prevention, Define the allowed deviation from the true value, n represents the training examples; The R-squared value in SVR is used to measure the extent to which the independent variable explains the variance of the dependent variable. It is a key indicator for evaluating the model fit in regression analysis. The corresponding expression is: （12） in, is the sum of squares of the errors between the observed and predicted values, is the sum of squares and measures the total variance of the dependent variable.

2. The GEE-based environmental monitoring and phenological correlation processing method according to claim 1, characterized in that: Also includes: The support vector regression algorithm is used to analyze and compare the meteorological data with the output environmental factors to verify the analysis results corresponding to the phenological correlation, wherein the meteorological data includes the monthly average values ​​of temperature, humidity and precipitation in the area to be studied.

3. The method for environmental monitoring and phenological correlation processing based on GEE according to claim 1, characterized in that: Obtain environmental factors of the area to be studied, including: The remote sensing images corresponding to the environmental factors are subjected to data preprocessing, wherein the preprocessing includes masking clouds and shadows in the remote sensing images, wherein functions cloudMaskL457 and maskL8sr are used to mask low-quality observations, and bit masks are created using pixel quality assessment bands to identify shadows and clouds through bit operations.

4. A GEE-based environmental monitoring and phenological correlation processing system, characterized in that: The method for environmental monitoring and phenological correlation processing based on GEE as claimed in any one of claims 1 to 3 comprises: An environmental factor acquisition module is used to acquire environmental factors of the area to be studied, and to perform environmental monitoring on the GEE platform using Landsat 8 remote sensing data based on the environmental factors to obtain a spatiotemporal map of the environmental factors, wherein the environmental factors include meteorological data and surface environmental variables, the surface environmental variables include soil adjusted vegetation index, surface temperature and soil moisture index, and the spatiotemporal map includes a time series map and a time line map corresponding to the surface environmental variables respectively; A phenological index determination module, used to determine a temporal soil-adjusted vegetation phenological index from a maximum vegetation index and a minimum vegetation index according to the spatiotemporal graph, and use the temporal soil-adjusted vegetation phenological index as a phenological measurement value for phenological correlation analysis; The phenological correlation analysis module is used to analyze the phenological correlation between the phenological measurement value and the index value corresponding to the surface environmental variable based on the support vector regression algorithm, wherein the index value includes the minimum value, the maximum value, the average value and the difference value.

Images