CN114723330A - Vegetation change influence factor evaluation method based on structural equation model - Google Patents

Abstract

The invention discloses an evaluation method for vegetation change influencing factors based on a structural equation model. The method mainly includes: (1) collecting NDVI data and NDVI variation impact factor dataset; (2) preprocessing the data; (3) establishing a conceptual structural equation model and inputting the processed variables into the model; (4) ) to calculate the goodness-of-fit index of the model; (5) to correct the model; (6) to finally obtain the influence mechanism of vegetation change, and analyze the results. The invention realizes quantitative evaluation of the influencing factors of vegetation change, and effectively quantifies the direct, indirect and total influence of each influencing factor on vegetation change, so as to provide scientific basis for regional ecological environment protection and decision-making management.

Description

An Evaluation Method of Influencing Factors of Vegetation Change Based on Structural Equation Model

technical field

The invention relates to the field of vegetation, in particular to a method for evaluating influence factors of vegetation change based on a structural equation model.

Background technique

Vegetation growth is strongly affected by topography, climate change and human activities, and changes in vegetation cover will affect the global or regional ecological environment. Understanding the temporal and spatial changes of regional vegetation and exploring its response to various influencing factors can provide theoretical support for regional ecological environmental protection.

Remote sensing technology is an effective means to realize long-time series, large-scale spatial-scale vegetation monitoring. The normalized difference vegetation index (NDVI) can accurately reflect the vegetation coverage on the surface and is the most common index for evaluating vegetation growth. Its annual maximum value can effectively reflect the best annual vegetation growth. Most studies focus on the direct effects of independent variables on dependent variables, ignoring indirect effects, resulting in biased results.

Therefore, quantifying the direct and indirect effects of each influencing factor on vegetation cover change and obtaining the total influence of each influencing factor on vegetation change is of great significance to regional ecological environment protection and decision-making.

SUMMARY OF THE INVENTION

Aiming at the deficiencies of the prior art, the present invention proposes an evaluation method for vegetation change influencing factors based on a structural equation model, so as to obtain the direct, indirect and total influences of each influencing factor on vegetation change.

The purpose of the present invention can be realized by the following technical solutions: a kind of evaluation method of vegetation change influence factor based on structural equation model, comprises the following steps:

S1. Data collection: collect and download NDVI data; collect and download a dataset of influencing factors affecting NDVI changes;

S2. Data preprocessing: including one or more of data splicing and cropping, data projection conversion, data reclassification, data resampling, and sampling point extraction;

S3, according to the factor selected in step S1, through step S2 processing, take it as the input variable of the structural equation model;

S4, establish a conceptual structural equation model, input the variables obtained in step S3 into the model, and calculate various goodness-of-fit indicators of the model;

S5. According to the fit criteria of the goodness-of-fit index, determine whether the goodness-of-fit indicators of the models calculated in step S4 meet the requirements; if so, determine the main influence of the NDVI change according to the significance and the size of the fitted standardized path coefficient According to the sign of the standardized path coefficient, the influence direction of each influencing factor on the change of NDVI is judged, and the direct and indirect influence of each influencing factor on the change of NDVI is judged according to the path relationship from the independent variable to the dependent variable, and the influence of each influencing factor on the change of NDVI is obtained. On the contrary, there are two main ways to revise the model: 1) revise the model through the model revision value, 2) revise the model by deleting insignificant variables and paths, and repeat step S4;

S6. Analyze the influencing factors of NDVI change according to the finally obtained structural equation model, and obtain the quantitative analysis result of the influencing factors of NDVI change.

Further, the impact factor data set described in step S1 includes: terrain factor, temperature factor, precipitation factor, and human activity factor; wherein the terrain factor is measured by two indicators of elevation and slope, and the human activity factor is measured by night lights, population density and land. Use three indicators to measure.

Further, in step S3, according to the impact factor data set selected in step S1, through the processing of step S2, the obtained input variables are specifically:

Extract the elevation value of each sampling point;

Extract the slope value of each sampling point;

Extract the slope of the temperature change at each sampling point;

Extract the slope of the precipitation change at each sampling point;

Extract the slope of night light changes at each sampling point;

Extract the slope of the population density change at each sampling point;

According to whether the land use type has changed, extract the value of each sampling point;

Extract the slope of the NDVI change at each sampling point.

Further, in the step S5, when the model goodness of fit index calculated in the step S4 meets the requirements, the path relationship between the topography, air temperature, precipitation, human activities, and NDVI changes will be obtained. The size of the coefficient determines the main influencing factors of NDVI change, and the influence direction of each influencing factor on NDVI change is judged according to the sign of the standardized path coefficient, that is, positive influence and negative influence, and each influencing factor is obtained according to the path relationship between independent variables and dependent variables. The influence mechanism of NDVI change was determined, and the direct and indirect influence of each influencing factor on NDVI change was judged, and the total influence of each influencing factor on NDVI change was obtained.

Further, in the step S5, modifying the model according to the model correction value and deleting insignificant variables and paths is specifically:

1) Arrange the model correction values in descending order, and establish the correlation between variables in order, so as to correct the structural equation model;

2) According to the significance test results of each path, the insignificant variables or paths are deleted, so as to revise the model.

Beneficial effects of the present invention: the present invention selects the influencing factors of vegetation change, takes them as the input variables of the structural equation model through preprocessing, and uses the structural equation model to iteratively analyze each variable, and establishes in the order of the correction values from large to small. The correlation between variables and the deletion of insignificant variables and paths are used to correct the model, so as to obtain the influence degree, influence direction and influence mode of each influencing factor on vegetation change, and effectively quantify the direct and indirect effects of each influencing factor on vegetation change. The total impact of each influencing factor on vegetation changes can be obtained, and then a scientific basis for regional ecological environment construction and sustainable development can be provided.

Description of drawings

The present invention will be further described below in conjunction with the accompanying drawings.

Fig. 1 is the flow chart of the evaluation method of vegetation change influence factor based on structural equation model of the present invention;

2 is a schematic diagram of a conceptual structural equation model established in an embodiment of the present invention;

Fig. 3 is the initial structural equation model result diagram in the embodiment of the present invention;

FIG. 4 is a normalization path diagram of a final fitted structural equation model in an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

A method for evaluating factors affecting vegetation change based on a structural equation model, comprising the following steps:

S1. Data collection: collect and download NDVI data; collect and download a dataset of influencing factors affecting NDVI changes;

S2. Data preprocessing: including one or more of data splicing and cropping, data projection conversion, data reclassification, data resampling, and sampling point extraction;

S3, according to the factor selected in step S1, through step S2 processing, take it as the input variable of the structural equation model;

S4, establish a conceptual structural equation model, input the variables obtained in step S3 into the model, and calculate various goodness-of-fit indicators of the model;

S5. According to the fit criteria of the goodness-of-fit index, determine whether the goodness-of-fit indicators of the models calculated in step S4 meet the requirements; if so, determine the main influence of the NDVI change according to the significance and the size of the fitted standardized path coefficient According to the sign of the standardized path coefficient, the influence direction of each influencing factor on the change of NDVI is judged, and the direct and indirect influence of each influencing factor on the change of NDVI is judged according to the path relationship from the independent variable to the dependent variable, and the influence of each influencing factor on the change of NDVI is obtained. On the contrary, there are two main ways to revise the model: 1) revise the model through the model revision value, 2) revise the model by deleting insignificant variables and paths, and repeat step S4;

S6. Analyze the influencing factors of NDVI change according to the finally obtained structural equation model, and obtain the quantitative analysis result of the influencing factors of NDVI change.

Further, the impact factor data set described in step S1 includes: terrain factor, temperature factor, precipitation factor, and human activity factor; wherein the terrain factor is measured by two indicators of elevation and slope, and the human activity factor is measured by night lights, population density and land. Use three indicators to measure.

Further, in step S3, according to the impact factor data set selected in step S1, through the processing of step S2, the obtained input variables are specifically:

Extract the elevation value of each sampling point;

Extract the slope value of each sampling point;

Extract the slope of the temperature change at each sampling point;

Extract the slope of the precipitation change at each sampling point;

Extract the slope of night light changes at each sampling point;

Extract the slope of the population density change at each sampling point;

According to whether the land use type has changed, extract the value of each sampling point;

Extract the slope of the NDVI change at each sampling point.

Further, in the step S5, when the model goodness of fit index calculated in the step S4 meets the requirements, the path relationship between the topography, air temperature, precipitation, human activities, and NDVI changes will be obtained. The size of the coefficient determines the main influencing factors of NDVI change, and the influence direction of each influencing factor on NDVI change is judged according to the sign of the standardized path coefficient, that is, positive influence and negative influence, and each influencing factor is obtained according to the path relationship between independent variables and dependent variables. The influence mechanism of NDVI change was determined, and the direct and indirect influence of each influencing factor on NDVI change was judged, and the total influence of each influencing factor on NDVI change was obtained.

Further, in the step S5, modifying the model according to the model correction value and deleting insignificant variables and paths is specifically:

1) Arrange the model correction values in descending order, and establish the correlation between variables in order, so as to correct the structural equation model;

2) According to the significance test results of each path, the insignificant variables or paths are deleted, so as to revise the model.

In a specific embodiment, the embodiment of the present invention takes Anhui Province as a case area, and takes 2000-2018 as a research period, and provides a structural equation model-based evaluation method for vegetation change influencing factors, which specifically includes the following steps:

S1. Data collection: collect and download NDVI data and the impact factor dataset of NDVI changes;

Download MOD13Q1 NDVI and SRTM DEM data based on Google Earth Engine (GEE) cloud platform. The spatial resolution of MOD13Q1 NDVI data is 250m and the temporal resolution is 16d; the spatial resolution of SRTM DEM data is 30m.

The population density data comes from the world population density map published by WorldPop ( https://www.worldpop.org/ ) with a spatial resolution of 1km.

The nighttime light data are derived from an extended time series (2000-2018) of global similar NPP-VIIRS nighttime light data based on cross-sensor calibration, and the dataset has a spatial resolution of 500 m.

The land use data comes from the land use dataset of Anhui Province provided by the Resource and Environmental Science Data Center of the Chinese Academy of Sciences ( http://www.resdc.cn ), which includes cultivated land, forest land, grassland, water area, residential land, and unused land There are 6 primary classes and 25 secondary classes, with a spatial resolution of 1 km.

The climate data (annual mean temperature and annual precipitation) were obtained from the China Meteorological Data Network ( http://data.cma.cn/ ) 0.5°×0.5° grid point dataset (V2.0) of monthly surface temperature in China and the Chinese surface temperature Precipitation monthly value 0.5°×0.5° grid data set (V2.0). See Table 1 for data preparation:

Table 1. Data introduction table

S2. Data preprocessing: including one or more of data splicing and cropping, data projection conversion, data reclassification, data resampling, and sampling point extraction;

Extract elevation and slope information from DEM data;

All the factors required by the case were cropped with the case area as the boundary, and all data were projected to the WGS_1984_UTM_Zone_50N coordinate system in ArcGIS, and resampled to the same spatial resolution (250m) as NDVI based on the nearest neighbor method to ensure data availability.

Based on the univariate linear regression method, the slopes of the annual average temperature change, annual precipitation change, nighttime light change, population density change and NDVI change from 2000 to 2018 were calculated pixel by pixel.

The pixels with changes in land use types in the case area from 2000 to 2018 are recorded as 0, and the pixels that have not changed are recorded as 1.

In the case, the water body was removed, the center point of the 3km×3km grid was taken, and the missing values were removed, and a total of 14,789 sampling points were extracted.

S3, according to the factor selected in step S1, through step S2 processing, take it as the input variable of the structural equation model;

Extract the elevation value of each sampling point;

Extract the slope value of each sampling point;

Extract the slope of the annual mean temperature change at each sampling point;

Extract the slope of the annual precipitation change at each sampling point;

Extract the slope of night light changes at each sampling point;

Extract the slope of the population density change at each sampling point;

According to whether the land use type has changed, extract the value of each sampling point;

Extract the slope of the NDVI change at each sampling point.

S4, establish a conceptual structural equation model, input the variables obtained in step S3 into the model, and calculate various goodness-of-fit indicators of the model;

In this embodiment, the present invention firstly establishes a conceptual structural equation model according to the NDVI variation and impact factor data selected in the above step S1, as shown in FIG. 2, and uses each factor obtained in the above step S3 as the input of the structural equation model The variables are iteratively analyzed, and various goodness-of-fit indicators of the model are obtained through calculation.

S5. According to the fit criteria of the goodness-of-fit index, determine whether the goodness-of-fit indicators of the models calculated in step S4 meet the requirements; if so, determine the main influence of the NDVI change according to the significance and the size of the fitted standardized path coefficient According to the sign of the standardized path coefficient, the influence direction of each influencing factor on the change of NDVI is judged, and the direct and indirect influence of each influencing factor on the change of NDVI is judged according to the path relationship from the independent variable to the dependent variable, and the influence of each influencing factor on the change of NDVI is obtained. The total impact of ; on the contrary, the model is revised, mainly in two ways:

1) Correct the model by the model correction value,

2) Delete insignificant variables and paths to correct the model, and repeat step S4;

In this embodiment, the present invention firstly compares each of the goodness-of-fit indexes obtained in step S4 with the goodness-of-fit index adaptation standards according to the goodness-of-fit index adaptation standard, and determines the model calculated in step S4 Whether the goodness of fit index meets the requirements.

If the various goodness-of-fit indicators of the structural equation model meet the fit criteria of the goodness-of-fit indicators, observe the final structural equation model, and judge the main impact of NDVI changes based on the significance and the size of the fitted standardized path coefficients According to the sign of the standardized path coefficient, the influence direction of each influencing factor on the change of NDVI is judged, and the direct and indirect influence of each influencing factor on the change of NDVI is judged according to the path relationship from the independent variable to the dependent variable, and the influence of each influencing factor on the change of NDVI is obtained. total impact.

If the various goodness-of-fit indicators of the structural equation model do not meet the adaptation criteria of the goodness-of-fit indicators, the model needs to be revised. There are mainly two ways: 1) The model is revised by the model correction value, Specifically, the model correction values are arranged in descending order, and a correlation is established between the corresponding two variables in order, or the variable with a large correction value is deleted. 2) Modify the model by deleting insignificant variables and paths, specifically deleting insignificant paths according to the saliency results of each path in the model. Repeat step S4;

In this embodiment, the present invention obtains the initial structural equation model estimation result as shown in FIG. 3 according to the conceptual structural equation model shown in FIG. 2 that was initially designed, and various goodness-of-fit indexes are shown in Table 2 shown.

Table 2. Results of various goodness-of-fit indicators of the initial structural equation model

In Table 2, χ2/df is the ratio of chi-square degrees of freedom, RMSEA is the root mean square of approximate error, SRMR is the normalized root mean square residual error, CFI is the comparative fit index, GFI is the goodness of fit index, and IFI is the correction The standard fit index of .

According to the goodness-of-fit indicators of the initial structural equation model obtained in Table 2, it can be seen that the model's χ2/df=17.483 (>3) does not meet the model's adaptation standard, so the model needs to be revised.

The invention corrects the model by establishing a correlation relationship according to the model correction value from large to small, deleting variables with large correction values or deleting insignificant variables and paths.

The correction coefficients of the initial model are shown in Table 3. According to the aforementioned method of correcting the model, a correlation can be established between e7 and e8 with the largest correction value, or the variable corresponding to e7 can be deleted.

Table 3. Correction values of initial structural equation model

The significant results of the path coefficients of the initial structural equation model are shown in Table 4. It can be seen from Table 4 that the paths between NDVI changes and annual mean temperature changes, and the paths between annual mean temperature changes and topography are not significant.

Table 4. Significance of path coefficients of initial structural equation model

Further, in order to improve the goodness of fit of the model, the present invention firstly deletes the annual average temperature variation variable in combination with the aforementioned model correction value results. The goodness-of-fit indicators of the revised model are shown in Table 5.

Table 5. Results of various goodness-of-fit indicators of the model after the first revision

Further, it can be seen from Table 5 that the model’s χ2/df=5.758<17.483, the fitting effect of the model has been improved to a certain extent, but χ2/df is still >3, which does not meet the model’s fitting criteria, so it is necessary to further analyze the model. Make corrections.

The correction values of the model after the first correction are shown in Table 6. According to the aforementioned model correction method, the present invention establishes the correlation in the order of the model correction values from large to small.

Table 6. Corrected values of the model after the first correction

Further, as can be seen from Table 6, first establish a correlation between e6 and e8 with the largest model correction value. The goodness-of-fit indicators of the revised model are shown in Table 7.

Table 7. Results of various goodness-of-fit indicators of the model after the second revision

It can be seen from Table 7 that the goodness of fit index of the model after the second revision has been further improved, but it still does not meet the fitting standard of the model fitting index, so the model needs to be further revised.

After the model is revised several times according to the aforementioned model revision method, various goodness-of-fit indicators of the final fitted structural equation model are obtained as shown in Table 8.

Table 8. Results of various goodness-of-fit indicators of the final fitted structural equation model

It can be seen from Table 8 that the various goodness-of-fit indicators of the structural equation model after multiple revisions all meet the model's adaptation criteria.

The standardized path coefficient and significance of the final fitted structural equation model are shown in Table 9, where the size of the standardized path coefficient represents the strength of the influence of the independent variable on the dependent variable, and the symbol represents the direction of the influence of the independent variable on the dependent variable. .

Table 9. Standardized path coefficients and significance of the final fitted structural equation model

It can be seen from Table 9 that each path of the model has passed the 0.001 level confidence test. Among them, the terrain not only has a significant negative impact on the changes of human activities (-0.14), but also has a significant positive impact on the annual precipitation changes. Changes in human activities and annual precipitation had a significant negative impact on NDVI changes (-0.47 and -0.04); topography had a significant positive impact on NDVI changes (0.06).

S6. Analyze the influencing factors of NDVI change according to the finally obtained structural equation model, and obtain the quantitative analysis result of the influencing factors of NDVI change.

In this embodiment, the present invention obtains the final fitted structural equation model according to the fitting result in step S5 as shown in FIG. 4 .

According to Figure 4, topography not only has a direct impact on the change of NDVI in Anhui Province, but also has an indirect impact, namely: 1) it has a significant positive impact on the change of NDVI; 2) it has a significant positive impact on the change of NDVI by inhibiting the change of human activities; 3) It has a significant negative impact on NDVI changes by promoting annual precipitation changes. Moreover, both elevation and slope can better measure the impact of terrain on NDVI changes. In addition, changes in human activities and changes in annual precipitation only have a direct impact on changes in NDVI. Among them, changes in human activities have a significant negative impact on changes in NDVI, and changes in nighttime lighting can better measure changes in human activities. Changes have significant negative effects.

Combining the direct and indirect effects of each influencing factor on the change of NDVI, the total influence of each influencing factor on the change of NDVI is obtained, and the results are shown in Table 10.

Table 10. The total impact of each influencing factor on the change of NDVI

From Table 10, it can be seen that the change of human activities has the greatest impact on the change of NDVI and is the main influencing factor of the change of NDVI. -0.04.

The invention uses the structural equation model to quantitatively analyze the influencing factors of vegetation change, and obtains the direct, indirect and total influence of each influencing factor on vegetation change, that is, the influence direction, influence degree and influence mechanism of each influence factor on vegetation change.

The above-mentioned embodiments are only to help readers understand the principles of the present invention, and are not intended to limit the present invention. The processing of data and the selection of research areas in the present invention can be determined according to requirements. On the premise of not departing from the spirit and scope of the present invention, the present invention will also have various changes and improvements, and these changes and improvements should all fall within the protection scope of the present invention.

Claims (5)

1. a vegetation change influencing factor evaluation method based on structural equation model, is characterized in that, comprises the following steps: S1. Data collection: collect and download NDVI data; collect and download a dataset of influencing factors affecting NDVI changes, where NDVI is the normalized vegetation index; S2. Data preprocessing: including several items in data splicing and cropping, data projection conversion, data reclassification, data resampling, and sampling point extraction; S3, according to the NDVI change influencing factor selected in step S1, process through step S2, and use it as the input variable of the structural equation model; S4, establish a conceptual structural equation model, input the variables obtained in step S3 into the model, and calculate various goodness-of-fit indicators of the model; S5. According to the fit criteria of the goodness-of-fit index, determine whether the goodness-of-fit indicators of the models calculated in step S4 meet the requirements; if so, determine the main influence of the NDVI change according to the significance and the size of the fitted standardized path coefficient According to the sign of the standardized path coefficient, the influence direction of each influencing factor on the change of NDVI is judged, and the direct and indirect influence of each influencing factor on the change of NDVI is judged according to the path relationship from the independent variable to the dependent variable, and the influence of each influencing factor on the change of NDVI is obtained. On the contrary, there are two main ways to revise the model: 1) revise the model through the model revision value, 2) revise the model by deleting insignificant variables and paths, and repeat step S4; S6. Analyze the influencing factors of NDVI change according to the finally obtained structural equation model, and obtain the quantitative analysis result of the influencing factors of NDVI change. 2. The method for evaluating vegetation change influence factors based on structural equation model according to claim 1, wherein the influence factor data set in the step S1 comprises: terrain factors, air temperature factors, precipitation factors, human Activity factor; the terrain factor is measured by two indicators of elevation and slope, and the human activity factor is measured by three indicators: night light, population density and land use. 3. a kind of vegetation change influence factor evaluation method based on structural equation model according to claim 1, is characterized in that, in described step S3, according to the influence factor data set selected in step S1, through step S2 processing, obtain The input variables are specifically: Extract the elevation value of each sampling point; Extract the slope value of each sampling point; Extract the slope of the temperature change at each sampling point; Extract the slope of the precipitation change at each sampling point; Extract the slope of night light changes at each sampling point; Extract the slope of the population density change at each sampling point; According to whether the land use type has changed, extract the value of each sampling point; Extract the slope of the NDVI change at each sampling point. 4. a kind of vegetation change influence factor evaluation method based on structural equation model according to claim 1, is characterized in that, in described step S5, when the model goodness-of-fit index calculated in step S4 meets the requirement, will The path relationship among topography, air temperature, precipitation, human activities, and NDVI changes, the main influencing factors and directions of NDVI changes, the influencing mechanisms of each influencing factor on NDVI changes, and the total impact of each influencing factor on NDVI changes were obtained. 5. a kind of vegetation change influence factor evaluation method based on structural equation model according to claim 1, is characterized in that, in described step S5, this model is revised specifically as: 1) Arrange the model correction values in descending order, and establish the correlation between variables in order, so as to correct the structural equation model; 2) According to the significance test results of each path, the insignificant variables or paths are deleted, so as to revise the model.

Images