CN116739133A - Regional reed NDVI pattern simulation prediction method based on natural-social dual-drive analysis - Google Patents

Abstract

The invention relates to the technical field of vegetation pattern simulation and regulation. Specifically, it is a regional reed NDVI pattern simulation and prediction method based on natural-society dual-driven analysis, taking into account policy planning, social development, land evolution and other social factors on the distribution of the macro-reed pattern. It also considers the impact of natural conditions on the distribution of reed micro-patterns. By combining the driving effects of natural factors and social factors, it simultaneously realizes the simulation of multi-scenario programs, overcoming the traditional method that only considers the natural driving effects and has limited simulation options. limitation.

Description

A simulation prediction of regional reed NDVI pattern based on natural-society dual-driven analysis Measurement method

Technical field

The invention relates to the technical field of vegetation pattern simulation and regulation, specifically a method for simulating and predicting the regional reed NDVI pattern based on natural-society dual-driven analysis.

Background technique

Vegetation is an important component of the ecosystem and plays an important role in maintaining its ecological balance and environmental quality. For ecosystems with serious ecological environment damage, vegetation pattern regulation is an effective way to repair their damaged structures and functions. However, due to the dual interference of climate change and human activities, vegetation growth has been damaged and distribution patterns have been encroached upon, which not only endangers the survival of the vegetation itself, but also threatens the health of the ecological environment.

Due to its long growth time, wide distribution range, and strong ability to purify water bodies, reeds are the main targets for vegetation pattern regulation. The Normalized Difference Vegetation Index (NDVI) can eliminate the interference of external factors and enhance the response ability to vegetation changes. It is often used to reflect the growth status and distribution of regional vegetation. Therefore, accurately identifying the influence of natural and social factors on the NDVI pattern of reed, and simulating and predicting the distribution of NDVI pattern of reed under changing environments are of great significance for vegetation pattern regulation and ecological environment restoration.

Currently, the reed NDVI pattern simulation methods commonly used in existing technologies at home and abroad include mathematical statistical analysis method and spatial analysis model method:

The mathematical statistical analysis method is based on the relationship between NDVI and natural driving factors of temperature and precipitation, and uses multiple linear regression methods to predict changes in NDVI. Although it involves driving mechanisms, it only considers the driving effects of natural factors and does not involve the influence of human factors in socio-economic development and development and utilization activities; and this method is only applicable to small sample size data at certain points and limited scenarios. For research, it would take a lot of time to conduct multi-scenario simulations of NDVI at all points in the regional space, and the efficiency is low.

The spatial analysis model method is based on the NDVI spatial transfer matrix to predict changes in the regional NDVI pattern. However, it only predicts based on the historical spatial change patterns of NDVI and does not involve the driving mechanism, making it difficult to predict the evolution of NDVI in a changing environment.

In summary, the current simulation and prediction method of regional reed NDVI pattern has the following limitations:

1. It does not involve the driving mechanism of NDVI changes, or only considers the driving mechanism of natural factors, and does not involve the driving role of socio-economic factors.

2. The calculation efficiency is low and it is impossible to quickly and efficiently conduct a comprehensive simulation and prediction of the reed NDVI pattern in the entire region.

3. Due to imperfect consideration of driving mechanisms and low computational efficiency, current simulation methods can consider limited changing environments and policy scenarios, and cannot simulate multiple composite change scenarios that combine socioeconomic development and natural environment changes.

Therefore, in order to solve the above problems, this application proposes a regional reed NDVI pattern simulation and prediction method based on natural-society dual-driven analysis to fill the gap in this field. It not only considers the impact of policy planning on the macroscopic pattern distribution of reed, but also considers the natural The influence of conditions on the distribution of reed micro-patterns can be realized to simulate the multi-scenario regional reed NDVI pattern taking into account the dual effects of socio-economic development and natural climate change.

Contents of the invention

The purpose of this invention is to overcome the shortcomings of the existing technology, provide a regional reed NDVI pattern simulation and prediction method based on natural-society dual-driven analysis, fill the gaps in this field, and not only consider the impact of policy planning on the macroscopic pattern distribution of reed, but also Taking into account the impact of natural conditions on the distribution of micro-patterns of reed, a multi-scenario regional reed NDVI pattern simulation is realized that takes into account the dual impacts of socioeconomic development and natural climate change. In order to achieve the above objectives, the present invention provides a regional reed NDVI pattern simulation and prediction method based on natural-society dual drive analysis, which includes the following steps:

S1, collect regional land use pattern data in historical periods, identify land use types and the areas of each category; collect social, economic, hydrological, climate and environmental data in the same historical period;

S2, divide the entire socio-economic-ecological environment composite system into four modules: socio-economic development, water demand, water quality and land type area demand, qualitatively analyze the interaction relationship between variables within each module and between modules, and map the area changes of different land types Causal feedback loop diagram;

S3. Quantitatively analyze the interaction equations between variables, determine the parameter values of each variable, and establish a regional land area demand system dynamics model to reflect natural factors through the interaction mechanism and feedback relationship between the socioeconomic system and the ecological environment system. and the dual driving effects of social factors on land type area changes, especially reed area changes;

S4, use historical data to calibrate the parameters of the system dynamics model constructed in S3, and test the validity of the model prediction results;

S5: Design different change scenarios based on future policy planning and climate change characteristics, use system dynamics models to predict the area needs of various types of areas in the region under different scenarios in the future, and carry out research based on the impact of different social development and natural change scenarios on area changes of land types. Multi-scenario predictive analysis;

S6. Collect data on driving factors of regional land use pattern changes. Factor data includes terrain elements, meteorological elements, soil conditions and location;

S7. Based on the historical land use pattern and pattern change driving factor data, combined with relevant research and expert suggestions, determine the total probability of each species appearing in specific patch units, and comprehensively consider the driving mechanisms of natural and social conditions on changes in the spatial pattern of land types. At the same time, combined with the spatial change laws of land use pattern;

S8, based on the spatial change rules of land use pattern explored in S7, use the FLUS model to build a regional land use spatial pattern distribution simulation model;

S9, use historical data to test the simulation accuracy of the FLUS model built in S8, analyze the difference between the simulated pattern and the actual pattern, and test the validity of the FLUS model simulation results;

S10, use the FLUS model, combined with the area requirements of each category under different scenarios obtained in S5, to simulate the spatial pattern distribution of regional land use under different scenarios, and on this basis, further extract the spatial pattern distribution of reeds under each scenario;

S11, collect historical NDVI and NDVI change driving factor data;

S12, after eliminating outliers in the data, the effective data set is divided into training data set, test data set and other data sets;

S13, analyze the correlation between historical NDVI data and NDVI change driving factor data, and determine the driving mechanism of regional NDVI changes;

S14, based on historical NDVI data and NDVI change driving factor data, use the random forest algorithm to build an NDVI prediction model, determine the optimal parameter values of the model, use the training data set to train the model, and obtain a regional reed NDVI simulation random forest model;

S15, use the test data set and other data sets to test the simulation performance of the NDVI simulated random forest model built in S14, and evaluate whether the prediction accuracy of the model meets the requirements;

S16, based on the spatial pattern distribution of reeds under different scenarios obtained from S10, and based on the future temperature, precipitation, and groundwater depth parameters designed in S5, use the random forest model to simulate and predict the regional future reed NDVI pattern distribution under different scenarios. In the traditional Based on the natural driving mechanism, it combines the influence of policy planning, social development, and social factors of land evolution, reflecting the dual driving effects of nature and society on changes in the pattern of reed NDVI.

In S9, if the Kappa coefficient between the simulated pattern and the actual pattern is greater than 0.6, the model simulation effect is significant and has passed the validity test. The calculation formula of the Kappa coefficient is:

Formula 1:

Formula 2:

p ₀ is the overall classification accuracy, which refers to the proportion of correctly classified land type samples in the total sample size; a _i is the area of the i-th type of land in the actual pattern; b _i is the area of the i-th type of land in the simulated pattern; n is the total sample size.

In S15, the determination coefficient R ² is selected as the evaluation index of the model simulation accuracy. If R ² is greater than 0.5, the prediction accuracy of the model is considered to meet the requirements and has passed the validity test. The calculation formula of the determination coefficient R ² is:

Formula three:

X _i and Y _i are the actual value and simulated value of the i-th sample respectively, n is the number of samples, is the mean of the actual values. Compared with the prior art, the present invention has the following beneficial effects:

1. Because the system dynamics model can clearly reflect the complex, nonlinear, dynamic action mechanisms and feedback loops between the socioeconomic system and the ecological environment system, the FLUS model can combine the spatial vegetation pattern based on the simulation results of the system dynamics model. Change patterns, simulate the distribution of regional reed spatial patterns under different natural changes and social development scenarios, successfully combine the driving effects of natural factors and social factors, and simultaneously realize the simulation of multi-scenario programs, overcoming the traditional method that only considers natural driving forces The role and limitations of the simulation program are limited.

2. Because the random forest model has the advantage of efficiently processing large sample size data, it can accurately simulate the NDVI value of each point in the region based on the simulation of the spatial pattern of reeds and the natural driving mechanism of NDVI changes, and obtain different natural and policy results. The distribution of regional reed NDVI patterns under the scenario overcomes the limitations of traditional methods such as long calculation time, low simulation efficiency, and inability to simulate the entire region.

Description of drawings

Figure 1 is a schematic flow chart of the simulation prediction method of the present invention.

Figure 2 is a schematic diagram of the socio-economic development module of the causal feedback loop of area changes of different land types in Baiyangdian according to the embodiment of the present invention.

Figure 3 is a schematic diagram of the water demand module of the causal feedback loop for area changes in different land types in Baiyangdian according to the embodiment of the present invention.

Figure 4 is a schematic diagram of the water quality module of the causal feedback loop for area changes in different land types in Baiyangdian according to the embodiment of the present invention.

Figure 5 is a schematic diagram of the land area demand module of the cause and effect feedback loop for area changes of different land types in Baiyangdian according to the embodiment of the present invention.

Figure 6 is a schematic diagram of the overall structure of the causal feedback loop model of Baiyangdian area changes in different land types according to the embodiment of the present invention.

Figure 7 is a schematic diagram of the actual land use pattern and simulated pattern of Baiyangdian in 2015 according to the embodiment of the present invention.

Figure 8 is a schematic diagram of the actual land use pattern and simulated pattern of Baiyangdian in 2020 according to the embodiment of the present invention.

Figure 9 is a schematic diagram of the land use pattern distribution under multiple scenarios in Baiyangdian in 2025 according to the embodiment of the present invention.

Figure 10 is a schematic diagram of the distribution of reed NDVI under multiple scenarios in Baiyangdian in 2025 according to the embodiment of the present invention.

Detailed ways

The present invention will now be further described with reference to the accompanying drawings.

Referring to Figures 1 to 6, the present invention provides a regional reed NDVI pattern simulation and prediction method based on natural-society dual-driven analysis. The Baiyangdian Wetland reed NDVI pattern simulation is used as an implementation case to demonstrate the operation process and implementation effect of the present invention:

Baiyangdian is located in the center of Hebei Province. It is the largest shallow lake wetland in the North China Plain. The ravines throughout the lake divide the entire lake into multiple interconnected lakes of different sizes, forming reed terraces, garden terraces, village buildings, and open water. Integrated characteristic land use pattern. As a typical vegetation in Baiyangdian wetland, reed is the main regulatory object for vegetation pattern restoration. Therefore, it is necessary to simulate the changes and distribution of reed NDVI pattern in Baiyangdian under different development scenarios.

As shown in Figure 1, applying the present invention to the simulation of the NDVI pattern of reeds in the Baiyangdian Wetland requires following three parts and a total of 16 operating steps, as follows:

1. Simulation of land area demand under multiple scenarios in Baiyangdian:

(1) Collect the land use pattern data of Baiyangdian in 2010, 2015 and 2020. The data type is 30m raster data. It can be seen that the main land use types of Baiyangdian are divided into four categories: construction land, cultivated land, reed terrace fields and water areas, and statistics for each The area of each category in each period; collect the social (population, birth rate, population reduction rate, per capita food demand, per capita water consumption) and economic (total GDP, industrial GDP, water consumption per 10,000 yuan of industrial GDP) Baiyangdian from 2010 to 2015 ), hydrology (lake water volume, water entering the lake, water leaving the lake, leakage, recycled water, water replenishment, ecological water consumption), climate (temperature, precipitation), environment (pollution entering the lake, pollution removal by reeds quantity) data.

(2) Based on the characteristics of Baiyangdian, divide the entire Baiyangdian system into four modules: socioeconomic development, water demand, water quality, and land type area demand. Qualitatively analyze the interaction relationship between variables within each module and between modules, and draw the areas of different land types. Change cause and effect feedback loop diagram, as shown in Figure 2 to Figure 6;

(3) Quantitatively analyze the relationship equations between each variable, and determine the parameter values of each variable, as shown in Table 1:

Table 1 Baiyangdian Land Type Area Demand System Dynamics Model Relationship Expression and Parameter Settings

(4) Taking 2010 as the starting year and supporting data in 2010 as the initial data, other parameter settings are shown in Table 1. Set the model time step to 1 year to simulate the area demand of each category in 2015 and 2020. At the same time Compared with the actual values, the results are shown in Table 2. The validity test results show that the prediction relative error range of the system dynamics model constructed by the present invention is 0.03% to 5.59%, all within the acceptable error range of 10%, so the prediction performance of the model is good.

Table 2 Simulation results and verification of Baiyangdian land area demand system dynamics model

(5) In terms of socioeconomic development, two scenarios are set: current development and ecological migration. In the current development scenario, the socioeconomic parameters remain at the 2010-2020 level. In the ecological migration scenario, it is assumed that 60% of Baiyangdian's population moves out every year. In terms of climate change, with reference to the CMIP6 climate change model, two models, SSP1-2.6 (sustainable development) and SSP5-8.5 (fossil fuel-driven development), were selected. Under the SSP1-2.6 model, the annual temperature increase is 0.01°C and the rainfall increases by 0.3mm. ; Under SSP5-8.5 mode, the annual temperature increase is 0.06℃ and the rainfall increases by 2.3mm. In terms of ecological environment protection, with reference to the Baiyangdian ecological environment management and protection plan, two water level modes are set, 7.3m (ecological water level) and 7.5m (flood control safety water level). Based on the above scenario changes, the changes in area demand of various types in Baiyangdian in 2025 are simulated. The results are shown in Tables 3 and 4.

Table 3 Area requirements of various categories in 2025 under different climate and water level scenarios under the current development background (×10 ⁷ m ² )

Table 4 Area requirements of various categories in 2025 under different climate and water level scenarios under the background of ecological migration (×10 ⁷ m ² )

2. Simulation of reed spatial pattern distribution under multiple scenarios in Baiyangdian:

(6) Collect data on the driving factors of land use pattern changes in Baiyangdian from 2010 to 2020, including DEM data and extract slope and aspect distribution, temperature and rainfall data, sand, silt, clay, and organic carbon content data on the soil surface, Anxin County road network vector map and calculation of distance from roads, Baiyangdian river network map and calculation of distance from rivers; the above data need to be processed into 30m raster data, the unified geographical coordinate system is GCS_WGS_1984, and the projected coordinate system is WGS_1984_UTM_zone_50N.

(7) Based on the historical land use pattern and pattern change driving factor data, use the artificial neural network module in the FLUS model, set the sampling rate to 30/1000, and the number of hidden layers to 30, to calculate the various locations in each specific patch unit in the space. The probability of class occurrence; at the same time, combined with relevant research and expert suggestions, the neighborhood factor parameters and land class conversion cost matrix of Baiyangdian land class conversion are determined, as shown in Table 5 and Table 6 respectively.

Table 5 Neighborhood factor parameters of the FLUS model of Baiyangdian land use pattern change

Table 6 Baiyangdian land conversion cost matrix

(8) Taking the 2010 land use pattern as the base period data, use the land type occurrence probability and neighborhood factor parameters calculated in step (7), and set the maximum number of iterations, acceleration factor, and parallel running threads of the model to 300 and 0.1 respectively. and 1. Use the cellular automaton module in the FLUS model to simulate the spatial distribution of land use patterns in Baiyangdian in 2015 and 2020. The results are shown in Figures 7 and 8.

(9) Based on the actual land use pattern of Baiyangdian in 2015 and 2020 and the land use pattern simulated in step (8), the Kappa coefficient was calculated to verify the simulation accuracy of the FLUS model. The results showed that the Kappa of the overall simulation result of the land use pattern in 2015 was The coefficient is 0.6338, among which the Kappa coefficients of the reed and water simulation results are 0.7260 and 0.6844 respectively; the Kappa coefficient of the overall simulation results in 2020 is 0.7163, among which the Kappa coefficients of the reed and water simulation results are 0.7856 and 0.7009 respectively. The overall Kappa coefficient of land use pattern is greater than 0.6, and the Kappa coefficient of reed is greater than 0.7, indicating that the FLUS model simulation results are significant.

(10) Use the FLUS model constructed in step (8), use the actual land use pattern in 2020 as the base period data, and use the 2025 Baiyangdian area demand of various types under the multiple scenarios simulated in step (5) as the simulation target to simulate each scenario The distribution of land use pattern in Baiyangdian in 2025 is shown in Figure 9. On this basis, the spatial pattern distribution of reeds under each scenario is further extracted.

3. Simulation of reed NDVI pattern distribution under multiple scenarios in Baiyangdian:

(11) Collect Baiyangdian reed NDVI data from 2010 to 2019 in May each year; at the same time, collect NDVI change driving factor data from 2010 to 2019, that is, average temperature data from January to April each year, and cumulative rainfall data from January to April. and groundwater depth data, where the groundwater depth is calculated by the following formula:

GWD＝H _L -H _W

Among them, GWD is the groundwater depth of the reed terrace (m); H _L is the land surface elevation (m), that is, DEM data; H _W is the average water level of Baiyangdian Lake (m).

(12) Unify a total of 10 periods of NDVI, temperature, precipitation and groundwater depth data from 2010 to 2019 into a comprehensive data set, eliminate outliers generated in the original data and interpolation calculation process, and use random sampling to select valid data pairs. 3% of the data pairs are extracted to form the modeling data set, and the other data pairs that are not extracted form other data sets. For the modeling data set, 70% of the data are randomly selected to form the training data set, and the remaining 30% of the data form the test data set for the construction and testing of the machine learning model.

(13) Based on the effective data set preprocessed in step (12), analyze the correlation between reed NDVI data and temperature, precipitation, and groundwater depth data, and verify the driving effect of each driving factor on the changes in reed NDVI. The results show that for For Baiyangdian reed, temperature, precipitation and groundwater depth are the key driving factors affecting its NDVI changes.

(14) Based on the training data set preprocessed in step (12), use the random forest algorithm to construct a Baiyangdian reed NDVI change simulation model, and use the tuneRF traversal function to search for optimal algorithm parameter values. The results show that when the value of mtry is 7 The model has the highest prediction accuracy, that is, setting the mtry parameter to 7, training the model again using the training data set, and obtaining the optimal Baiyangdian reed NDVI simulation model.

(15) Use the test data set and other data sets to test the simulation performance of the NDVI simulated random forest model constructed in step (14), and calculate the coefficient of determination (R ² ) of the model in the three data sets. The results show that during training The R ² of the set, test set and other set models are 0.9008, 0.5233 and 0.5164 respectively, all greater than 0.5, indicating that the model prediction performance is good and the simulation results are valid.

(16) Based on the spatial pattern distribution of reeds under different scenarios obtained in step (10), use the optimal Baiyangdian reed NDVI simulation model constructed in step (14) to predict the NDVI pattern distribution of reeds in Baiyangdian in 2025 under each scenario. The results are as follows As shown in Figure 10.

It can be seen from the above examples that the regional reed NDVI pattern simulation and prediction method proposed by the present invention based on natural-society dual-driven analysis focuses on the impact of socioeconomic factors on the spatial pattern distribution of reeds during the construction of the system dynamics model and FLUS model. The driving role of natural elements is also taken into account; while the random forest model modeling inherits the traditional method of considering the natural driving mechanism, absorbing the advantages of traditional technology, and expanding on traditional technology. . At the same time, because the system dynamics model has good multi-scenario simulation advantages and the random forest model has good big data processing capabilities, the method proposed by the present invention can analyze the NDVI in each patch unit based on the spatial pattern distribution of reeds. The value is simulated and predicted, thereby realizing the simulation of the NDVI pattern of reed in the entire region, breaking through the limitation of traditional methods that can only simulate the NDVI value of a few specific point units.

The above are only the preferred embodiments of the present invention, and are only used to help understand the method and its core idea of the present application. The protection scope of the present invention is not limited to the above-mentioned embodiments. All technical solutions that fall under the ideas of the present invention belong to the present invention. scope of protection. It should be pointed out that for those of ordinary skill in the art, several improvements and modifications may be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

The present invention overall solves the problems in the prior art that do not involve a driving mechanism, that it is difficult to predict the evolution of NDVI under a changing environment, that there are few factors to consider, that the calculation efficiency is low, and that all points in the target area cannot be studied. Combining the driving effects of natural factors and social factors, it simultaneously realizes the simulation of multi-scenario scenarios, overcoming the limitations of traditional methods that only consider the natural driving effects and limited simulation options.

Claims (3)

1. The regional reed NDVI pattern simulation prediction method based on natural-social dual-drive analysis is characterized by comprising the following steps of:

s1, collecting land utilization pattern data in a historical period of an area, and identifying land utilization types and areas of various land types; collecting social, economic, hydrological, climate and environmental data of the same historical period;

s2, dividing the whole socioeconomic-ecological environment composite system into four modules of socioeconomic development, water demand, water quality and land area demand, qualitatively analyzing interaction relations of variables inside each module and among the modules, and drawing different land area change causal feedback loop diagrams;

s3, quantitatively analyzing an action relation equation among the variables, determining the parameter values of the variables, establishing a regional land area demand system dynamics model, and reflecting the double driving action of natural factors and social factors on land area change, especially reed area change through action mechanisms and feedback relations between a social economic system and an ecological environment system;

s4, parameter calibration is carried out on the system dynamics model constructed in the S3 by utilizing historical data, and validity of a model prediction result is checked;

s5, different change scenes are designed according to future policy planning and climate change characteristics, the system dynamics model is used for predicting the area requirements of each type of land under different future scenes, and multi-scene prediction analysis is carried out based on the influence of different social development and natural change scenes on the area change of the land;

s6, collecting regional land utilization pattern change driving factor data, wherein the factor data comprise terrain elements, meteorological elements, soil conditions and location positions;

s7, determining the total probability of each land type in a specific plaque unit based on historical land utilization pattern and pattern change driving factor data, combining related research and expert advice, comprehensively considering a driving mechanism of the natural condition and the social condition on the land type space pattern change, and simultaneously combining the space change rule of the land utilization pattern;

s8, constructing a regional land utilization space pattern distribution simulation model by using an FLUS model based on the land utilization pattern space change rule explored in the S7;

s9, performing simulation precision test on the FLUS model constructed in the step S8 by using historical data, analyzing the difference between a simulation pattern and an actual pattern, and testing the effectiveness of a simulation result of the FLUS model;

s10, simulating land utilization space pattern distribution of areas under different scenes by using an FLUS model and combining the area requirements of each scene under the different scenes obtained in the S5, and further extracting reed space pattern distribution conditions under each scene on the basis;

s11, collecting historical NDVI and NDVI change driving factor data;

s12, dividing the effective data set into a training data set, a test data set and other data sets after eliminating abnormal values in the data;

s13, analyzing the correlation between the historical NDVI data and the NDVI change driving factor data, and determining a driving mechanism of the NDVI change of the area;

s14, based on the historical NDVI data and the NDVI change driving factor data, constructing an NDVI prediction model by using a random forest algorithm, determining an optimal parameter value of the model, and training the model by using a training data set to obtain a regional reed NDVI simulation random forest model;

s15, testing the simulation performance of the NDVI simulation random forest model constructed in the S14 by using the test data set and other data sets, and evaluating whether the prediction precision of the model meets the requirement;

s16, based on reed space pattern distribution under different scenes obtained from the S10, based on future temperature, precipitation and underground water burial depth parameters designed in the S5, simulating and predicting future reed NDVI pattern distribution conditions of areas under different scenes by using the random forest model, combining the effects of policy planning, social development and land evolution social elements on the basis of a traditional natural driving mechanism, and reflecting the double driving effect of natural-society on reed NDVI pattern change.

2. The method for simulating and predicting the natural-society double-drive analysis-based area reed NDVI pattern according to claim 1, wherein in S9, if the Kappa coefficient between the simulated pattern and the actual pattern is greater than 0.6, the model simulation effect is significant, and the validity test is passed, and the calculation formula of the Kappa coefficient is:

equation one:

formula II:

the p is ₀ For the overall classification accuracy, the ratio of the accurately classified land sample size to the total sample size is pointed out; the a _i The area of the i-th land in the actual pattern; said b _i The area of the i-th land in the simulation pattern; and n is the total sample size.

3. The method for simulating and predicting the NDVI pattern of a regional reed based on natural-society dual drive analysis as claimed in claim 1, wherein in S15, a decision coefficient R is selected ² As an evaluation index of the model simulation accuracy, if R ² If the prediction accuracy of the model is greater than 0.5, the prediction accuracy of the model is considered to meet the requirement, and the validity test is passed, and the decision coefficient R is determined ² The calculation formula of (2) is as follows:

and (3) a formula III:

the X is _i And Y _i The actual value and the analog value of the ith sample are respectively, n is the number of samples, andis the average of the actual values.