CN106485360A - Segmental society's prediction of economic indexes method and system based on overall noctilucence remote sensing - Google Patents

Abstract

The present invention provides a method and system for predicting local socioeconomic indicators based on global luminous remote sensing, including clipping images and extracting the total amount of night light intensity values in the target range, and extracting socioeconomic indicators in local areas; based on the assumption of linear consistency, judging the global area Whether there is a linear correlation between the total amount of nighttime light intensity values and the total amount of nighttime light intensity values in a local area, and between the total amount of nighttime light intensity values in a local area and the socioeconomic indicators in a local area, satisfying the preset condition threshold ; When it is satisfied, build a quantitative inversion model of local socio-economic indicators based on global night light remote sensing. According to the total amount of night light intensity values in the global region in the year to be predicted, predict and analyze based on the inversion model, and obtain the socio-economic conditions of the local region in the corresponding year. The index prediction results can complete the prediction of the local social and economic development situation, and the prediction results can be obtained automatically and in time, which has important market value.

Description

Prediction method and system of local social and economic indicators based on global luminous remote sensing

technical field

The invention belongs to the application field of remote sensing geosciences, and relates to a method and system for quantitative inversion prediction of local socio-economic indicators based on global luminous remote sensing.

Background technique

The Operational Line System (OLS) sensor carried by the Defense Meteorological Satellite Program (DMSP) of the United States has a strong ability to amplify photoelectricity because it uses an optical multiplier tube (PMT) at night. Using its photoelectric amplification characteristics, it can detect city lights, fires and even extremely low-visibility glows, so it has the ability to clearly capture the footprints of human activities in a dark background, and has become a unique tool that can monitor human economic activities. The data. The available DMSP/OLS is the long-term data from 1992 to 2013, which ensures the comparability of night light remote sensing data between regions and years. For this reason, DMSP/OLS night light remote sensing data are widely used in large-scale urban spatial expansion research, economic and population estimation, urban electricity consumption and energy consumption analysis, carbon emissions and light pollution and other environmental issues assessment.

However, the spatial resolution of DMSP/OLS data is 2.7km, which is obtained by averaging five higher-resolution sensor data on the satellite. The application and research of OLS data can only be concentrated in larger areas at or above the city level, while the application and research of small areas based on DMSP/OLS night light data are rarely involved. At the same time, the existing large-area modeling based on DMSP/OLS data has not been rigorously proved theoretically, and lacks the rationality and scientificity of theoretically demonstrating the model construction.

Contents of the invention

The purpose of the present invention is to provide a quantitative inversion and forecasting technical solution for local socio-economic indicators based on global luminous remote sensing in view of the shortcomings of the existing DMSP/OLS data in the application range (small scale) and model construction theory demonstration.

The technical solution adopted in the present invention provides a local socio-economic index prediction method based on global luminous remote sensing, comprising the following steps:

Step 1, using the local area and the global area as the target range, cut out the night light remote sensing image of the target range, and extract the total nighttime light intensity value SOL of the target range for several years;

Step 2, extracting the socio-economic indicators SEI _Local for several years in the local area;

Step 3. Based on the assumption of linear consistency, determine the difference between the total night light intensity value SOL _Global in the global area and the total night light intensity value SOL _Local in the local area, and the difference between the total night light intensity value SOL _Loca in the local area and SOL Local. Whether there is a linear correlation between the socioeconomic indicators SEI _Local in the local area that meets the threshold of the preset condition, including the following sub-steps,

Step 3.1, determine whether there is a linear correlation between SOL _Global and SOL _Local that meets the threshold of the preset condition, and the calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total night light in the local area, (SOL _Global ) _i is the total night light in the i-th year in the global area, is the expectation of the total amount of lights at night in the global area; N represents the total number of years, and the value of i is 1,2,...,N;

Step 3.2, judge whether there is a linear correlation between SOL _Local and SEI _Local that meets the threshold of the preset condition, and the calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total amount of light at night in a local area, (SEI _Local ) _i is the socio-economic index value in the i-th year in the local area, is the expectation of the socio-economic index value in the local area;

Step 4, when the linear correlation conditions in step 3.1 and step 3.2 are established, by constructing a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing, it is used to predict the local socio-economic development trend, including the following sub-steps,

Step 4.1, establish the following ideal model and perform parameter estimation,

SEI _Local ＝A+B×SOL _Global

where A and B are the estimated model parameters;

In step 4.2, according to the model parameters estimated in step 4.1, a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing is constructed as follows,

SEI _Local ＝A+B×SOL _Global +ζ,ζ～N(0,σ ² )

In the formula, ζ is the overall random error after disturbance, N(0,σ ² ) is a normal distribution with a mean of 0 and a variance of σ;

Step 5: According to the total amount of nighttime light intensity values in the global region in the year to be predicted, and based on the quantitative inversion model of local socio-economic indicators based on global luminous remote sensing constructed in step 4, carry out prediction and analysis, and obtain the forecast of socio-economic indicators in the local region in the corresponding year As a result, the local socio-economic development situation estimation is completed.

Moreover, the threshold value of the preset condition is 0.8, and when 0.8<|ρ|≤1, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local are highly correlated.

Moreover, in step 1 and step 5, the realization of extracting the total amount of light intensity values at night includes the following sub-steps,

Step 1.1, convert the geographic projection coordinate system of DMSP/OLS luminous remote sensing data into Lambert equal-area projection to obtain DMSP/OLS raster image;

Step 1.2, according to the vector boundary layer of the target range, cut out the DMSP/OLS raster image map of the target range;

Among them, n represents the number of gray levels of the pixel, DN _i represents the brightness value of the i-th level pixel in the area, M _i represents the total number of i-th brightness level pixels in the area, and i ranges from 0 to n-1.

Moreover, the prediction result obtained in step 5, its 1-α prediction interval is:

Among them, α is the significance level, SOL ₀ is the SOL of the global region that satisfies the correlation condition, that is, the total value of nighttime light intensity values in the global region in the year to be predicted, and SEI ₀ is the observed value at SOL=SOL ₀ , representing the value to be predicted Annual local area socio-economic indicators; is the estimate of SEI ₀ , representing the prediction results of socioeconomic indicators in local areas in the year to be predicted, and δ(SOL ₀ ) is the uncertainty of the predicted value.

The present invention provides a local socio-economic index prediction system based on global luminous remote sensing, including the following modules:

The first module is used to take the local area and the global area as the target range, cut out the night light remote sensing image of the target range, and extract the total nighttime light intensity value SOL of the target range for several years;

The second module is used to extract the socio-economic indicators SEI _Local for several years in a local area;

The third module is used to determine the difference between the total night light intensity value SOL _Global in the global area and the total night light intensity value SOL _Local in the local area based on the assumption of linear consistency, and the total night light intensity value in the local area Whether there is a linear correlation between SOL _Loca and the socioeconomic index SEI _Local of the local area that meets the threshold of the preset condition, including the following units,

Unit 3.1 is used to judge whether there is a linear correlation between SOL _Global and SOL _Local that meets the threshold of the preset condition. The calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total night light in the local area, (SOL _Global ) _i is the total night light in the i-th year in the global area, is the expectation of the total amount of lights at night in the global area; N represents the total number of years, and the value of i is 1,2,...,N;

Unit 3.2 is used to judge whether there is a linear correlation between SOL _Local and SEI _Local that meets the threshold of the preset condition. The calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total amount of light at night in a local area, (SEI _Local ) _i is the socio-economic index value in the i-th year in the local area, is the expectation of the socio-economic index value in the local area;

The fourth module is used to predict the local socio-economic development trend by constructing a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing when units 3.1 and 3.2 determine that the linear correlation condition is established, including the following units, unit 4.1 , establish the following ideal model and perform parameter estimation,

SEI _Local ＝A+B×SOL _Global

where A and B are the estimated model parameters;

In unit 4.2, according to the model parameters estimated in unit 4.1, the quantitative inversion model of local socio-economic indicators based on global night light remote sensing is constructed as follows,

SEI _Local ＝A+B×SOL _Global +ζ,ζ～N(0,σ ² )

In the formula, ζ is the overall random error after disturbance, N(0,σ ² ) is a normal distribution with a mean of 0 and a variance of σ;

The fifth module is used to predict and analyze the global nighttime light intensity value of the global region in the year to be predicted, based on the quantitative inversion model of local socio-economic indicators based on the global nighttime light remote sensing constructed by the fourth module, and obtain the local region's value of the corresponding year. The prediction results of socio-economic indicators are used to complete the estimation of local socio-economic development trends.

Moreover, the threshold value of the preset condition is 0.8, and when 0.8<|ρ|≤1, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local are highly correlated.

Moreover, in the first module and the fifth module, the realization of extracting the total amount of light intensity values at night includes the following units,

Unit 1.1 is used to convert the geographic projection coordinate system of DMSP/OLS luminous remote sensing data into Lambert equal-area projection to obtain DMSP/OLS raster images;

Unit 1.2 is used to crop the DMSP/OLS raster image of the target range according to the vector boundary layer of the target range;

Among them, n represents the number of gray levels of the pixel, DN _i represents the brightness value of the i-th level pixel in the area, M _i represents the total number of i-th brightness level pixels in the area, and i ranges from 0 to n-1.

Moreover, the prediction result obtained by the fifth module, its 1-α prediction interval is:

Among them, α is the significance level, SOL ₀ is the SOL of the global region that satisfies the correlation condition, that is, the total value of nighttime light intensity values in the global region in the year to be predicted, and SEI ₀ is the observed value at SOL=SOL ₀ , representing the value to be predicted Annual local area socio-economic indicators; is the estimate of SEI ₀ , representing the prediction results of socioeconomic indicators in local areas in the year to be predicted, and δ(SOL ₀ ) is the uncertainty of the predicted value.

The beneficial effects of the technical solution provided by the present invention are as follows: Aiming at the deficiencies in the existing night light remote sensing image research methods and contents, a new global to local quantitative inversion model prediction method based on night light data is proposed, and from Theoretically, the transferability of global night light remote sensing images to local night light remote sensing images is deduced and proved. On this basis, the proposed ideal quantitative inversion model is optimized, the RANSAC algorithm can also be used to improve the data quality, and the uncertainty of the model estimation can be quantitatively analyzed, which expands the application range of luminous remote sensing data, and contributes to the study of geographical The study of global to local issues in science and other disciplines provides a new way of thinking, estimating social and economic indicators is more objective, and avoids the interference of human factors. Different from the traditional statistical method, which requires a large amount of data from various aspects and takes up a large amount of manpower and material resources, the technical solution provided by the present invention can automatically obtain the estimated results in time according to the total amount of night light intensity values in the global area in the current year. significant market value.

Description of drawings

Fig. 1 is a flow chart of an embodiment of the present invention.

Fig. 2 is a schematic diagram of night light remote sensing images of Hubei Province (global area) and Wuhan City (local area) according to an embodiment of the present invention.

Fig. 3 is a schematic diagram of constructing a linear regression prediction model from SOL _Hubei to Wuhan GDP _Wuhan before and after removing outliers in the embodiment of the present invention, wherein Fig. 3a is a schematic diagram of constructing a linear regression prediction model before removing outliers, and Fig. 3b is a schematic diagram of removing outliers Schematic diagram of the construction of the final linear regression prediction model.

Figure 4 is a schematic diagram of the comparison of GDP _Wuhan and _{SOL Hubei} prediction intervals _before and after removing _outliers in the embodiment of the present invention _. Schematic of prediction intervals.

detailed description

In order to better understand the technical solution of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.

Referring to Fig. 1, the embodiment of the present invention takes the quantitative inversion of Wuhan's GDP based on night light remote sensing in Hubei Province as an example, for the total amount of light intensity at night (Sum of Lights, SOL) and socioeconomic indicators (Socio-Economic Indicators, SEI), The total value of nighttime light intensity in the global area is recorded as SOL _Global , and the corresponding experimental example is SOL _Hubei ; the total value of nighttime light intensity in the local area is recorded as SOL _Local , and the corresponding experimental example is SOL _Wuhan ; The socio-economic index is marked as SEI _Local , and the corresponding experimental example is GDP _Wuhan .

The light intensity value at night in the embodiment is derived from DMSP/OLS night light remote sensing data, and other data may also be used during specific implementation, not limited to DMSP/OLS.

In the embodiment, the specific implementation steps of the local socio-economic index prediction method based on global luminous remote sensing are as follows:

Step 1, using the local area and the global area as the target range, use the extraction tool in ArcGIS to cut out the night light remote sensing image of the target range, and obtain the number of target ranges through the calculation formula of the total value of light intensity at night (Sum of Lights, SOL). Annual SOL.

The specific implementation includes the following sub-steps,

Step 1.1, convert the geographic projection coordinate system of DMSP/OLS luminous remote sensing data into Lambert equal-area projection to obtain DMSP/OLS raster image;

Step 1.2, according to the vector boundary layer of the target range, cut out the DMSP/OLS raster image map of the target range from the result obtained in step 1;

Step 1.3, through the statistical summation method, the calculation method of the total SOL value of the nighttime light intensity value in the target range is obtained as follows:

Among them, n represents the number of grayscale levels of the pixel, and the DMSP/OLS luminous data records a 6-bit grayscale image, so the grayscale level is 2 to the 6th power, that is, 64 levels. DN _i represents the brightness value of the i-th level pixel in the area, M _i represents the total number of i-th brightness level pixels in the area, i ranges from 0 to n-1, so the embodiment takes from 0 to 63, and 1 can be used for calculation to 63.

During specific implementation, the target range is the designated area, and the total value of nighttime light intensity can be extracted for the local area or the global area as the target range, respectively, as required. See Figure 2 for night light remote sensing images of Hubei Province (global area) and Wuhan City (local area).

Step 2, determine the indicators (Socio-Economic Indicators, SEI) to measure the socio-economic situation of the local area, such as GDP or population, etc., the implementation method can be obtained by consulting the national or regional statistical yearbook of the relevant indicators, and get the socio-economic indicators of the local area for several years Index SEI _Local ;

According to formula (1) and relevant statistical yearbooks, the SOL of Hubei Province and Wuhan in 1998, 2000, 2003, 2005, 2006, 2008, 2009, 2010, 2011, 2012 and other 10 years were calculated. The city's social and economic indicators GDP, as shown in Table 1 below,

Table 1 Hubei Province SOL (SOL _Hubei ) and Wuhan City GDP (GDP _Wuhan ) (unit: ten thousand yuan)

Step 3, based on the assumption of linear consistency, according to the historical data obtained in steps 1 and 2, determine the difference between the total night light intensity value (SOL _Global ) in the global area and the SOL (SOL _Local ) in the local area, as well as the SOL (SOL Local ) in the local area Whether there is a strong linear correlation between SOL _Local ) and local regional socio-economic indicators (Socio-Economic Indicators, SEI) (SEI _Local ), including the following sub-steps,

Step 3.1, to determine whether there is a strong linear correlation between SOL _Global and SOL _Local , the calculation formula of the correlation coefficient is as follows:

In formula (2), (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total night light in the local area, obtained by taking the average of (SOL _Local ) _i in N years, (SOL _Global ) _i is the total night light in the i-th year in the global area, is the expectation of the total amount of lights at night in the global area (that is, the arithmetic mean); N represents the total number of years, and the value of i is 1, 2,...,N.

Step 3.2, judge whether there is a strong linear correlation between SOL _Local and SEI _Local , the calculation formula of the correlation coefficient is as follows:

In formula (3), SOL _Local is the total amount of night light in the local area, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total amount of light at night in a local area, (SEI _Local ) _i is the socio-economic index value in the i-th year in the local area, is the expectation (ie arithmetic mean) of the socio-economic index value in the local area.

The embodiment of the present invention proposes that when |ρ|=0, it indicates that SOL _Global and SOL _Local or SOL _Local and SEI _Local are completely irrelevant; when 0<|ρ|<0.3, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local No correlation; when 0.3<|ρ|≤0.5, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local are lowly correlated; when 0.5<|ρ|≤0.8, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local are significantly correlated Correlation; when 0.8<|ρ|≤1, it is considered that SOL _Global and SOL _Local or SOL _Local and SEI _Local are highly correlated.

In this embodiment, the correlation condition threshold is set to 0.8, and those skilled in the art can set it to other values during specific implementation. If the threshold value of the correlation condition is met, it is determined that the condition of strong linear correlation is established, and the step 4 can be entered. If it is not satisfied, it is not suitable for this method, and the process is stopped. According to formula (2) and formula (3), the calculated correlation coefficients between SOL _Hubei and SOL _Wuhan and SOL _Wuhan and GDP _Wuhan are 0.9748 and 0.0.9122 respectively, and both correlation coefficients satisfy ρ∈(0.8,1], The linear dependence conditions in step 3.1 and step 3.2 hold.

Step 4, when the linear correlation conditions in Step 3.1 and Step 3.2 are established, estimate the local socio-economic development trend by constructing a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing, including the following sub-steps,

Step 4.1, in order to build a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing, first construct an ideal model and perform parameter estimation:

According to the strong correlation between SOL _Global and SOL _Local and between SOL _Local and SEI _Local , the regression model can be constructed as follows:

SOL _Local =a _DN +b _DN ×SOL _Global (4)

SEI _Local ＝a _Local +b _Local ×SOL _Local (5)

In the formula, SOL _Global represents the SOL in the global area, SOL _Local represents the SOL in the local area, SEI _Local represents the SEI in the local area, a _DN and b _DN , a _Local and b _Local are coefficients.

Substitute formula (4) into formula (5) to get SEI _Local :

SEI _Local ＝a _Local +b _Local ×(a _DN +b _DN ×SOL _Global ) (6)

Further, let A=a _Local +b _Local ×a _DN , B=b _Local ×b _DN , so the ideal model is as follows:

SEI _Local ＝A+B×SOL _Global (7)

where A and B are estimated model parameters.

The principle of correlation equivalence between SEI _Local and SOL _Global and SEI _Local and SOL _Local is proved as follows:

In the formula, ρ is the correlation coefficient between the SEI of the local area and the SOL of the local area, and ρ' is the correlation coefficient between the SEI of the local area and the SOL of the global area. It can be seen that from the local SOL to the local SEI The correlation coefficient ρ of is equal to the correlation coefficient ρ' between the global SOL and the local SEI.

In specific implementation, when estimating the parameters of the linear regression prediction model, the existing RANSAC algorithm can be used, that is, firstly, a judgment criterion is formulated for the input data, and those data that are greatly different from the estimated parameters are iteratively eliminated, and then the correct ones are retained. The input data solves the linear regression predictive model. The judgment criterion requires that at least one set of data can be guaranteed to be all internal points of the model under a certain confidence probability in the M group of data, that is, the following relationship needs to be satisfied:

d＝1-(1-(1-θ) ^q ) ^M (9)

In the formula, θ represents the error rate of the data (that is, the proportion of outliers in the original data), q is the minimum amount of data required to calculate the model parameters, d is the confidence probability, and M is the number of sampling groups. According to the actual situation of the experimental data, the final model parameters are estimated by calculating the minimum sampling number by setting the confidence probability d of the experimental data and the error rate θ of the data.

For example, use the existing RANSAC algorithm program to input the 10-year SOL (SOL _Hubei ) and Wuhan GDP (GDP _Wuhan ) data into the RANSAC algorithm program. According to the formula (7), set the confidence probability setting d to 0.90 and the maximum The number of iterations is 45, and the seventh and eighth points are removed, that is, the data of 2009 and 2010, and the estimated parameters of the model are -3977.725 and 0.013, respectively.

Step 4.2, according to the model parameters estimated in step 4.1, construct a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing, as follows:

SEI _Local ＝A+B×SOL _Global +ζ,ζ～N(0,σ ² ) (10)

In the formula, SEI _Local is the SEI of the local area, SOL _Global is the SOL of the global area, A and B are parameters to be estimated, ζ is the overall random error after disturbance, N(0,σ ² ) is a normal distribution, and the mean is 0, and the variance is σ.

Considering the disturbance in the actual data, the noise on the slope and intercept can be considered comprehensively, and the GTLQTM disturbance model under the actual disturbance can be constructed. Therefore, the expressions of SEI _Local and SOL _Local can be written as:

In formulas (11) and (12), i is used to mark the i-th observation data, m and n are the denominators of the slope, e' and f' are the intercepts, ω _i ', μ _i ', η _i ' , λ _i ' are random disturbance items added to the slope and intercept respectively, and obey the normal distribution. The expressions of SEI _Local and SOL _Global are as follows:

Among them, a and b are the first-order polynomial coefficients corresponding to the slope and intercept.

Through substitution, under actual disturbance conditions, the quantitative inversion conversion expression of global to local nighttime light remote sensing images is as follows:

SEI _Local ＝A+B×SOL _Global +ζ (14)

In formula (14), SEI _Local is the SEI of the local area, SOL _Global is the SOL of the global area, A and B are parameters to be estimated, and ζ is the overall random error after disturbance. According to formula (14), the quantitative inversion model of local socio-economic indicators based on global night light remote sensing is obtained, such as formula (10).

In the embodiment, according to formula (10), the quantitative inversion model of local socio-economic indicators based on global luminous remote sensing is constructed after removing outliers, and is y=-3977.725+0.013×x, y corresponds to SEI _Local , and x corresponds to SOL _Global . The scatter plot before and after removing outliers is shown in Figure 3. According to the formula (10), the regression model y=-1598.1072+0.0091×x was constructed before removing outliers (as shown in Figure 3a), the coefficient of determination R ² =0.6453, and the regression model after removing outliers was y=-3977.7251+0.013 ×x (as shown in Figure 3b), the coefficient of determination R ² =0.8709, after removing outliers, the goodness of fit of the model is significantly improved.

Step 5: According to the total amount of nighttime light intensity values in the global region in the year to be predicted, and based on the quantitative inversion model of local socio-economic indicators based on global luminous remote sensing constructed in step 4, carry out prediction and analysis, and obtain the forecast of socio-economic indicators in the local region in the corresponding year As a result, the local socio-economic development situation estimation is completed.

The method of extracting the total amount of nighttime light intensity values in the global region in the year to be predicted is consistent with step 1.

In practice, if the traditional statistical method is used, a large amount of data from various aspects is required, and a large amount of human and material resources are occupied, and the statistical results of social and economic indicators are often obtained with a lag of several years. Utilizing the method provided by the present invention, the prediction result can be obtained automatically and in time according to the total amount of light intensity values at night in the global area in the current year.

The technical scheme of the present invention is used to predict and analyze the local socio-economic index quantitative inversion model based on global luminous remote sensing, and its 1-α prediction interval is:

In the formula, SOL ₀ is the SOL of the global area that satisfies the correlation condition, that is, the total amount of nighttime light intensity values in the global area in the year to be predicted, and SEI ₀ is the observed value at SOL=SOL ₀ , that is, the value of the local area in the year to be predicted Socio-economic indicators; is the estimate of SEI ₀ , that is, the prediction result of the socioeconomic index of the local area in the year to be predicted; δ(SOL ₀ ) is the uncertainty of the predicted value (valuation) SOL, preferably, the calculation is as follows,

in, is the standard deviation estimate, t obeys the student distribution, represents the distribution of students quantile, is the mean value of SOL in all years in the historical data, the sum of the squares of the total deviation N is the number of observations, that is, the total number of years of historical data, SOL _i is the total value of night light intensity values in the global region in the i-th year in historical data, the degree of freedom is N-2, and α is the significance level.

According to the formula (15), let α=0.05, the 95% confidence interval is obtained as shown in Fig. 4 . Comparing Figure 4a and Figure 4b, the 95% confidence interval band in Figure 4b is narrower than the confidence interval band in Figure 4a. Regression models are better able to explain the sample situation.

Secondly, compare the minimum, maximum, and mean uncertainty intervals of x before and after removing outliers in Table 2. Before removing outliers, the uncertainty interval at x _min = 372019 is (2376.96, 5512.57), and the change The magnitude is 7889.53, the uncertainty interval at x _max = 953738 is (1749.74, 9516.04), the range of change is 7766.3, the uncertainty interval at x _mean = 591793 is (47.7678, 7477.54), the range of change is 7429.77; abnormalities are removed The uncertainty interval at x _min = 372019 after the value is (2299.77, 3432.44), the range of change is 5732.21, the uncertainty interval at x _max = 953738 is (3518.76, 9283.2), the range of change is 5764.44, x _mean = The uncertainty interval at 569746 is (783.86, 6076.07) and the range of change is 5292.21. After removing the outliers, the uncertainty interval is significantly reduced. Taking the minimum value, maximum value, and mean value of x as an example, the uncertainty interval The reduction is roughly 2000.

Table 2 Comparison of the true value and predicted value of GDP in Wuhan before and after removing outliers

During specific implementation, the method provided by the present invention can realize the automatic operation process based on software technology, and can also realize the corresponding system in a modular manner. An embodiment of the present invention provides a local socio-economic index prediction system based on global luminous remote sensing, including the following modules:

The first module is used to take the local area and the global area as the target range, cut out the night light remote sensing image of the target range, and extract the total nighttime light intensity value SOL of the target range for several years;

The second module is used to extract the socio-economic indicators SEI _Local for several years in a local area;

The third module is used to determine the difference between the total night light intensity value SOL _Global in the global area and the total night light intensity value SOL _Local in the local area based on the assumption of linear consistency, and the total night light intensity value in the local area Whether there is a linear correlation between SOL _Loca and the socioeconomic index SEI _Local of the local area that meets the threshold of the preset condition, including the following units,

Unit 3.1 is used to judge whether there is a linear correlation between SOL _Global and SOL _Local that meets the threshold of the preset condition. The calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total night light in the local area, (SOL _Global ) _i is the total night light in the i-th year in the global area, is the expectation of the total amount of lights at night in the global area; N represents the total number of years, and the value of i is 1,2,...,N;

Unit 3.2 is used to judge whether there is a linear correlation between SOL _Local and SEI _Local that meets the threshold of the preset condition. The calculation formula of the correlation coefficient is as follows,

Among them, (SOL _Local ) _i is the total amount of night light in the i-th year in the local area, is the average value of the total amount of light at night in a local area, (SEI _Local ) _i is the socio-economic index value in the i-th year in the local area, is the expectation of the socio-economic index value in the local area;

The fourth module is used to predict the local socio-economic development trend by constructing a quantitative inversion model of local socio-economic indicators based on global luminous remote sensing when units 3.1 and 3.2 determine that the linear correlation condition is established, including the following units, unit 4.1 , establish the following ideal model and perform parameter estimation,

SEI _Local ＝A+B×SOL _Global

where A and B are the estimated model parameters;

In unit 4.2, according to the model parameters estimated in unit 4.1, the quantitative inversion model of local socio-economic indicators based on global night light remote sensing is constructed as follows,

SEI _Local ＝A+B×SOL _Global +ζ,ζ～N(0,σ ² )

In the formula, ζ is the overall random error after disturbance, N(0,σ ² ) is a normal distribution with a mean of 0 and a variance of σ;

The fifth module is used to predict and analyze the global nighttime light intensity value of the global region in the year to be predicted, based on the quantitative inversion model of local socio-economic indicators based on the global nighttime light remote sensing constructed by the fourth module, and obtain the local region's value of the corresponding year. The prediction results of socio-economic indicators are used to complete the estimation of local socio-economic development trends.

For the specific implementation of each module, reference may be made to the corresponding steps, which will not be described in detail in the present invention.

The above content is a further detailed description of the present invention in conjunction with the embodiments, and it cannot be assumed that the specific implementation of the present invention is limited to these descriptions. Those skilled in the art should understand that without departing from the conditions defined by the appended claims, various modifications can be made in the details, which should be regarded as belonging to the protection scope of the present invention.

Claims (8)

1. A local socioeconomic index prediction method based on global noctilucent remote sensing is characterized by comprising the following steps:

step 1, respectively taking a local area and a global area as target ranges, cutting out a noctilucent remote sensing image of the target range, and extracting the total SOL of nighttime light intensity values of the target range for a plurality of years;

step 2, extracting the social and economic indicators SEI of the local area for a plurality of years_Local；

Step 3, based on linear consistency assumption, judging global areaTotal amount of light intensity at night SOL_GlobalSum of night light intensity values SOL associated with local area_LocalAnd total amount of night light intensity values SOL of local area_LocaSocial and economic indicators SEI with local areas_LocalAnd whether there is a linear correlation that satisfies a preset condition threshold, includes the substeps of,

step 3.1, judging SOL_GlobalAnd SOL_LocalWhether linear correlation meeting the preset condition threshold exists or not, the correlation coefficient calculation formula is as follows,

$\mathrm{\&rho;}=\frac{\underset{i=1}{\overset{N}{\&Sigma;}}\&lsqb;{\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}}\&rsqb;\&CenterDot;\&lsqb;{\left({\mathrm{SOL}}_{Global}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Global}}\&rsqb;}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}})}^2\&CenterDot;\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Global}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Global}})}^2}},\mathrm{\&rho;}\&Element;\&lsqb;-1,1\&rsqb;$

wherein (SOL)_Local)_iIs the total light quantity in the local area at the ith year night,is the average value of the total amount of local area night light, (SOL)_Global)_iIs the light total amount in the ith year night in the global area,is the expectation of the total amount of night light in the global area; n represents the total number of years, and the value of i is 1,2, … and N;

step 3.2, judging SOL_LocalAnd SEI_LocalWhether linear correlation meeting the preset condition threshold exists or not, the correlation coefficient calculation formula is as follows,

$\mathrm{\&rho;}=\frac{\underset{i=1}{\overset{N}{\&Sigma;}}\&lsqb;{\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}}\&rsqb;\&CenterDot;\&lsqb;{\left({\mathrm{SEI}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SEI}}_{Local}}\&rsqb;}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}})}^2\&CenterDot;\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SEI}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SEI}}_{Local}})}^2}},\mathrm{\&rho;}\&Element;\&lsqb;-1,1\&rsqb;$

wherein (SOL)_Local)_iIs the total light quantity in the local area at the ith year night,is the average value of the total amount of local area night light, (SEI)_Local)_iIs the i-th year socioeconomic index value in the local area,the expectation of socioeconomic index values in local areas is obtained;

step 4, when the linear correlation condition in the step 3.1 and the step 3.2 is established, a local socioeconomic index quantitative inversion model based on global noctilucent remote sensing is constructed for predicting the development situation of the local socioeconomic, and the method comprises the following substeps,

step 4.1, the following ideal model is established and parameter estimation is carried out,

SEI_Local＝A+B×SOL_Global

wherein A and B are estimated model parameters;

step 4.2, according to the model parameters estimated in the step 4.1, a local socioeconomic index quantitative inversion model based on global noctilucent remote sensing is constructed as follows,

SEI_Local＝A+B×SOL_Global+ζ,ζ～N(0,σ²)

where ζ is the overall random error after perturbation, N (0, σ)²) Is normally distributed, with a mean of 0 and a variance of σ;

and step 5, according to the total night light intensity value of the global area of the year to be predicted, performing prediction analysis based on the local socioeconomic index quantitative inversion model based on the global noctilucent remote sensing and constructed in the step 4 to obtain the socioeconomic index prediction result of the local area of the corresponding year, and completing the prediction of the development situation of the local socioeconomic.

2. The local socioeconomic indicator prediction method based on global noctilucent remote sensing according to claim 1, characterized in that: the preset condition threshold value is 0.8, and when | rho | is less than or equal to 1 in the condition of 0.8, the SOL is considered to be_GlobalAnd SOL_LocalOr SOL_LocalAnd SEI_LocalAre highly correlated.

3. The local socioeconomic indicator prediction method based on global noctilucent remote sensing according to claim 1 or 2, characterized in that: in step 1 and step 5, the implementation of extracting the total amount of the night light intensity value comprises the following substeps,

step 1.1, converting a geographical projection coordinate system of DMSP/OLS noctilucent remote sensing data into Labert equal-area projection to obtain a DMSP/OLS raster image;

step 1.2, cutting out a DMSP/OLS raster image map of a target range according to a vector boundary image layer of the target range;

$SOL=\underset{i=1}{\overset{n-1}{\&Sigma;}}{\left(DN\right)}_i\&times;M_i$

wherein n represents the number of pixel gray levels, DN_iBrightness value, M, representing picture elements of the ith level in the area_iRepresenting the total number of picture elements of the ith brightness level in the area, i being from 0 to n-1.

4. The local socioeconomic indicator prediction method based on global noctilucent remote sensing according to claim 3, characterized in that: and 5, obtaining a prediction result, wherein the prediction interval of 1-alpha is as follows:

wherein α is the significance level, SOL₀SOL of global region for satisfying correlation condition, i.e. total night light intensity value of global region for year to be predicted, SEI₀Is SOL ═ SOL₀The observed value represents the social and economic index of the local area of the year to be predicted;is SEI₀(ii) an estimation of socio-economic indicators prediction results (SOL) representing a local area of the year to be predicted₀) Is the uncertainty of the predicted value.

5. A local socioeconomic index prediction system based on global noctilucent remote sensing is characterized by comprising the following modules:

the first module is used for cutting out a noctilucent remote sensing image of a target range by taking a local area and a global area as the target range respectively, and extracting the total SOL of nighttime light intensity values of the target range for a plurality of years;

a second module for extracting the social economic indicators SEI of the local area for several years_Local；

A third module for determining the total amount SOL of night light intensity values in the global area based on the assumption of linear consistency_GlobalSum of night light intensity values SOL associated with local area_LocalAnd total amount of night light intensity values SOL of local area_LocaSocial and economic indicators SEI with local areas_LocalAnd whether there is a linear correlation satisfying a preset condition threshold, includes the following units,

unit 3.1 for judging SOL_GlobalAnd SOL_LocalWhether linear correlation meeting the preset condition threshold exists or not, the correlation coefficient calculation formula is as follows,

$\mathrm{\&rho;}=\frac{\underset{i=1}{\overset{N}{\&Sigma;}}\&lsqb;{\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}}\&rsqb;\&CenterDot;\&lsqb;{\left({\mathrm{SOL}}_{Global}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Global}}\&rsqb;}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}})}^2\&CenterDot;\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Global}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Global}})}^2}},\mathrm{\&rho;}\&Element;\&lsqb;-1,1\&rsqb;$

wherein (SOL)_Local)_iIs the total light quantity in the local area at the ith year night,is the average value of the total amount of local area night light, (SOL)_Global)_iIs the light total amount in the ith year night in the global area,is the expectation of the total amount of night light in the global area; n represents the total number of years, and the value of i is 1,2, … and N;

unit 3.2 for judging SOL_LocalAnd SEI_LocalWhether linear correlation meeting the preset condition threshold exists or not, the correlation coefficient calculation formula is as follows,

$\mathrm{\&rho;}=\frac{\underset{i=1}{\overset{N}{\&Sigma;}}\&lsqb;{\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}}\&rsqb;\&CenterDot;\&lsqb;{\left({\mathrm{SEI}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SEI}}_{Local}}\&rsqb;}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SOL}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SOL}}_{Local}})}^2\&CenterDot;\underset{i=1}{\overset{N}{\&Sigma;}}{({\left({\mathrm{SEI}}_{Local}\right)}_i-\stackrel{\&OverBar;}{{\mathrm{SEI}}_{Local}})}^2}},\mathrm{\&rho;}\&Element;\&lsqb;-1,1\&rsqb;$

wherein (SOL)_Local)_iIs the total light quantity in the local area at the ith year night,is the average value of the total amount of local area night light, (SEI)_Local)_iIs the i-th year socioeconomic index value in the local area,the expectation of socioeconomic index values in local areas is obtained;

the fourth module is used for predicting the development situation of the local socioeconomic by constructing a local socioeconomic index quantitative inversion model based on global noctilucent remote sensing when the unit 3.1 and the unit 3.2 judge that the linear correlation condition is established, and comprises the following units,

unit 4.1, building the following ideal model and performing parameter estimation,

SEI_Local＝A+B×SOL_Global

wherein A and B are estimated model parameters;

a unit 4.2, according to the model parameters estimated by the unit 4.1, constructs a local social and economic index quantitative inversion model based on global noctilucent remote sensing as follows,

SEI_Local＝A+B×SOL_Global+ζ,ζ～N(0,σ²)

where ζ is the overall random error after perturbation, N (0, σ)²) Is normally distributed, with a mean of 0 and a variance of σ;

and the fifth module is used for carrying out prediction analysis on the basis of the local socioeconomic index quantitative inversion model based on the global noctilucent remote sensing and constructed by the fourth module according to the total night light intensity value of the global area of the year to be predicted to obtain the socioeconomic index prediction result of the local area of the corresponding year, and completing the prediction of the local socioeconomic development situation.

6. The local socioeconomic indicator prediction system based on global night light remote sensing of claim 5, wherein: the preset condition threshold value is 0.8, and when | rho | is less than or equal to 1 in the condition of 0.8, the SOL is considered to be_GlobalAnd SOL_LocalOr SOL_LocalAnd SEI_LocalAre highly correlated.

7. The local socioeconomic indicator prediction system based on global night-light remote sensing according to claim 5 or 6, characterized in that: in the first module and the fifth module, the implementation of extracting the total amount of the night light intensity value comprises the following units,

the unit 1.1 is used for converting a geographical projection coordinate system of the DMSP/OLS noctilucent remote sensing data into a Labert equal-area projection to obtain a DMSP/OLS raster image;

a unit 1.2, configured to cut out a DMSP/OLS raster image map of the target range according to the vector boundary layer of the target range;

$SOL=\underset{i=1}{\overset{n-1}{\&Sigma;}}{\left(DN\right)}_i\&times;M_i$

wherein n represents the number of pixel gray levels, DN_iBrightness value, M, representing picture elements of the ith level in the area_iRepresenting the total number of picture elements of the ith brightness level in the area, i being from 0 to n-1.

8. The local socioeconomic indicator prediction system based on global night light remote sensing of claim 7, wherein: the prediction interval of 1-alpha of the prediction result obtained by the fifth module is as follows:

wherein α is the significance level, SOL₀SOL of global region for satisfying correlation condition, i.e. year to be predictedGlobal area total night light intensity value, SEI₀Is SOL ═ SOL₀The observed value represents the social and economic index of the local area of the year to be predicted;is SEI₀(ii) an estimation of socio-economic indicators prediction results (SOL) representing a local area of the year to be predicted₀) Is the uncertainty of the predicted value.