Multi-source long time series night light data reconstruction method and system

Abstract

The application discloses a multi-source long-time sequence night light data reconstruction method and system, relates to the field of ecological remote sensing data reconstruction, and is based on night light images of a first-generation satellite DMSP / OLS and a second-generation satellite NPP / VIIRS to develop a method for reconstructing a set of long-time sequence night light data products. Since the satellite images of the two generations of satellites are obtained in the same year, the hyperbolic sine transformation method is used to fit the NPP / VIIRS in 2013 into the data form of the DMSP / OLS, an optimal fitting equation is obtained, and a set of long-time sequence data products from 1992 to 2021 is produced. The application can solve the fault problem between the two generations of light data DMSP / OLS and NPP / VIRS, and improve the accuracy of data reconstruction.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ecological remote sensing data reconstruction, in particular to a long time series night light data reconstruction method and system based on DMSP / OLS and NPP / VIIRS. BACKGROUND

[0002] Night light index has timeliness and economy, and can represent urbanization level from a macro perspective, which is an important indicator of economic development level. DMSP / OLS and NPP / VIIRS are both night light data from NOAA, but these two data sources have been affected by data faults for a long time, making it difficult to establish a continuous and consistent data set for long time series research. At present, the research on the consistency integration of the two light data includes: Li et al. proposed a mutual calibration model based on power function and Gaussian low-pass filter to integrate DMSP / OLS and NPP / VIIRS data, and explored the changes of light brightness in human settlements in Syria from 2011 to 2017, but this method does not have universal data availability. Ma et al. and Zhao et al. proposed a new integration model based on a logistic function model to simulate DMSP / OLS data after 2013 with NPP / VIIRS data. The accuracy of this type of model can reach more than 95%. Yu et al. integrated the Yangtze River Delta urban agglomeration night light data set from 2001 to 2019 based on this model, and explored the spatial and temporal heterogeneity of urbanization in the Yangtze River Delta urban agglomeration. These studies have solved the inconsistency problem of DMSP / OLS and NPP / VIIRS data in some specific regions, but the applicability of these methods in other regions is uncertain. SUMMARY

[0003] The purpose of the present application is to provide a multi-source long time series night light data reconstruction method and system, which can solve the fault problem between the two generations of light data DMSP / OLS and NPP / VIRS, and improve the accuracy of data reconstruction.

[0004] To achieve the above purpose, the present application provides the following scheme:

[0005] A multi-source long time series night light data reconstruction method, the method comprising:

[0006] obtaining DMSP / OLS night light data set and NPP / VIIRS night light data set of a target region and performing spatial resolution uniform preprocessing;

[0007] performing data mutual correction between sensors on the preprocessed DMSP / OLS night light data set;

[0008] performing data continuity correction on the DMSP / OLS night light data set after sensor correction.

[0009] continuity correction is performed on the pretreated NPP / VIIRS nighttime light data set;

[0010] applying an inverse hyperbolic sine transformation to data fitting between the corrected DMSP / OLS nighttime light data set and the corrected NPP / VIIRS nighttime light data set in an overlapping period data set, to obtain a data fitting model;

[0011] applying the non-overlapping period NPP / VIIRS nighttime light data set to the data fitting model to obtain a non-overlapping period fitted DMSP / OLS nighttime light data set; the overlapping period fitted DMSP / OLS nighttime light data set and the non-overlapping period fitted DMSP / OLS nighttime light data set constitute a complete reconstructed light data set;

[0012] integrating the reconstructed light data set and the DMSP / OLS nighttime light data set to obtain a multi-source integrated complete data.

[0013] Optionally, the DMSP / OLS nighttime light data set and the NPP / VIIRS nighttime light data set of a target region are obtained and spatial resolution is unified for pretreatment, specifically including:

[0014] the DMSP / OLS nighttime light data set of a first preset year segment and the NPP / VIIRS nighttime light data set of a second preset year segment are respectively reprojected onto an Albers equal-area projection of a preset region;

[0015] the reprojected DMSP / OLS nighttime light data set and the reprojected NPP / VIIRS nighttime light data set are respectively resampled based on a nearest neighbor method;

[0016] the resampled DMSP / OLS nighttime light data set and the resampled NPP / VIIRS nighttime light data set are respectively data cropped according to boundary data of the target region, to obtain the pretreated DMSP / OLS nighttime light data set and the pretreated NPP / VIIRS nighttime light data set.

[0017] Optionally, the pretreated DMSP / OLS nighttime light data set is mutually corrected between sensors, specifically including:

[0018] the pretreated DMSP / OLS nighttime light data set is calibrated based on a pseudo-invariant region sequential calibration method, to obtain a sensor-intercorrected DMSP / OLS nighttime light data set.

[0019] Optionally, the calibration of each period of data of each sensor corresponding to the pre-processed DMSP / OLS night light data set is based on a pseudo-invariant region sequential calibration method, and specifically includes:

[0020] determining a reference sensor from each sensor corresponding to the pre-processed DMSP / OLS night light data set;

[0021] sequentially calculating each period of sensor data of a non-reference sensor by using a one-dimensional quadratic non-linear regression equation to obtain the inter-sensor corrected DMSP / OLS night light data set.

[0022] Optionally, the inter-sensor corrected DMSP / OLS night light data set is subjected to data continuity correction, and specifically includes:

[0023] According to the inter-sensor corrected DMSP / OLS night light data set, the DMSP / OLS night light data set of the overlapping years of the sensors is subjected to data processing by using a pixel-by-pixel averaging method to obtain an average-processed DMSP / OLS night light data set;

[0024] According to the DMSP / OLS night light data set corresponding to the reference sensor as a reference, the average-processed DMSP / OLS night light data set is subjected to continuity calibration.

[0025] Optionally, according to the DMSP / OLS night light data set corresponding to the reference sensor as a reference, the average-processed DMSP / OLS night light data set is subjected to continuity calibration, and specifically includes:

[0026] The DMSP / OLS night light data set of the years before the year corresponding to the reference sensor is subjected to continuity calibration year by year by using a first formula; the first formula is:

[0027] The DMSP / OLS night light data set of the years after the year corresponding to the reference sensor is subjected to continuity calibration year by year by using a second formula; the second formula is:

[0028] Optionally, before the continuity correction of the pre-processed NPP / VIIRS night light data set, the following steps are further included:

[0029] The pre-processed NPP / VIIRS night light data set is subjected to a T0.3 mask method to screen out light pixels with light brightness above a first preset value and below a second preset value to obtain a noise-reduced NPP / VIIRS night light data set.

[0030] Optionally, the data fitting model is obtained by applying the inverse hyperbolic sine transformation to the overlapping period data set between the corrected DMSP / OLS night light data set and the corrected NPP / VIIRS night light data set, and specifically comprising:

[0031] determining the overlapping period data set between the corrected DMSP / OLS night light data set and the corrected NPP / VIIRS night light data set;

[0032] applying the inverse hyperbolic sine transformation to the NPP / VIIRS night light data set of the overlapping period;

[0033] fitting the NPP / VIIRS night light data set after the inverse hyperbolic sine transformation of the overlapping period and the DMSP / OLS night light data set of the overlapping period by using the sigmoid logistic function to obtain the data fitting model; the NPP / VIIRS night light data set after the inverse hyperbolic sine transformation of the overlapping period is the DMSP / OLS night light data set after the fitting of the overlapping period.

[0034] Optionally, the expression of the inverse hyperbolic sine transformation is:

[0035]

[0036] The expression of the sigmoid logistic function is:

[0037] The application also provides a multi-source long time series night light data reconstruction system, which comprises:

[0038] a data acquisition and preprocessing module for acquiring DMSP / OLS night light data set and NPP / VIIRS night light data set of a target area and performing spatial resolution uniform preprocessing;

[0039] a first correction module for mutually correcting data between different sensors for the preprocessed DMSP / OLS night light data set;

[0040] a second correction module for correcting data continuity for the DMSP / OLS night light data set after the inter-sensor correction;

[0041] a third correction module for correcting continuity for the preprocessed NPP / VIIRS night light data set;

[0042] a fitting module for applying the inverse hyperbolic sine transformation to the overlapping period data set between the corrected DMSP / OLS night light data set and the corrected NPP / VIIRS night light data set to perform data fitting and obtain a data fitting model.

[0043] a data reconstruction module configured to apply the data fitting model to the NPP / VIIRS nighttime light data set of the non-overlapping period to obtain a DMSP / OLS nighttime light data set of the non-overlapping period after fitting; and the DMSP / OLS nighttime light data set of the overlapping period after fitting and the DMSP / OLS nighttime light data set of the non-overlapping period after fitting constitute a complete reconstructed nighttime light data set;

[0044] a multi-source data integration module configured to integrate the reconstructed nighttime light data set and the DMSP / OLS nighttime light data set to obtain a complete data after multi-source integration.

[0045] According to the specific embodiments of the present application, the following technical effects are provided:

[0046] The present application provides a multi-source long time series nighttime light data reconstruction method and system, which fits DMSP / OLS through hyperbolic sine transformation processed NPP / VIIRS (IHS (NPP / VIIRS)), obtains the optimal fitting model, the model has high fitting accuracy pixel by pixel, can further integrate two generations of nighttime light data DMSP / OLS and NPP / VIRS, solves the fault problem between the two generations of light data DMSP / OLS and NPP / VIRS, and obtains a set of multi-source reconstructed and continuous nighttime light data set product. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0048] Figure 1 A multi-source long time series nighttime light data reconstruction method flow chart is provided for embodiment 1 of the present application;

[0049] Figure 2 A fitting result comparison chart based on different fitting methods for data fitting model is provided for embodiment 1 of the present application;

[0050] Figure 3 A 2021 Guangdong-Hong Kong-Macao Greater Bay Area nighttime light map in the complete data after multi-source integration obtained by the multi-source data integration module provided for embodiment 1 of the present application. DETAILED DESCRIPTION

[0051] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of the present application.

[0052] The purpose of the present application is to provide a multi-source long time series night light data reconstruction method and system, which solves the fault problem between two generations of light data DMSP / OLS and NPP / VIRS, and improves the accuracy of data reconstruction.

[0053] In order to make the above-mentioned purposes, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0054] Embodiment 1

[0055] As shown in the figure, the present embodiment provides a multi-source long time series night light data reconstruction method taking the Guangdong-Hong Kong-Macao Greater Bay Area as an example, which specifically includes the following steps: Figure 1

[0056] S1: Obtain the DMSP / OLS night light data set and NPP / VIIRS night light data set of the target area and perform spatial resolution uniform preprocessing.

[0057] Projection transformation, resampling and clipping of DMSP / OLS and NPP / VIIRS original light image data sets:

[0058] First, the original NPP / VIIRS monthly light image data from 2013 to 2021 is averaged annually to obtain the original annual NPP / VIIRS light image data set from 2013 to 2021 (a total of 9 remote sensing image maps); 34 pieces of F10-F18 annual original light image data set of DMSP / OLS from 1992 to 2013 and 9 pieces of original annual NPP / VIIRS light image data set from 2013 to 2021 are respectively re-projected onto the Albers equal-area projection suitable for the China region (central meridian 105°E, standard latitude 25°N, 47°N), and then resampled based on the nearest neighbor method to reduce the grid cell to 1km resolution, so that all light image data are unified in spatial resolution; finally, the Guangdong-Hong Kong-Macao Greater Bay Area administrative boundary grid data is used to clip all light image data to obtain the original DMSP / OLS and NPP / VIIRS light image data set of the Guangdong-Hong Kong-Macao Greater Bay Area.

[0059] Among them, step S1 is summarized as: ​

[0060] (1) The DMSP / OLS nighttime light dataset of the first preset year and the NPP / VIIRS nighttime light dataset of the second preset year are respectively reprojected onto the Albers equal area projection of the preset area.

[0061] (2) Based on the nearest neighbor method, pixel resampling was performed on the reprojected DMSP / OLS night light dataset and the reprojected NPP / VIIRS night light dataset.

[0062] (3) The resampled DMSP / OLS night light dataset and the resampled NPP / VIIRS night light dataset are respectively cropped according to the boundary data of the target area to obtain the preprocessed DMSP / OLS night light dataset and the preprocessed NPP / VIIRS night light dataset.

[0063] S2: Perform cross-calibration of data between sensors on the preprocessed DMSP / OLS nighttime light dataset.

[0064] Based on the sequential calibration method using pseudo-invariant regions, the data from each sensor in each period corresponding to the preprocessed DMSP / OLS nighttime light dataset are calibrated to obtain the sensor-corrected DMSP / OLS nighttime light dataset. Specifically:

[0065] A reference sensor is determined from each sensor corresponding to the preprocessed DMSP / OLS nighttime light dataset;

[0066] The sensor data of the non-benchmark sensor in each period are calculated sequentially using a univariate quadratic nonlinear regression equation to obtain the DMSP / OLS nighttime light dataset after inter-sensor correction.

[0067] In step S2, calibration is performed on 34 annual light image data (F10-F18) from DMSP / OLS from 1992 to 2013 across different sensors. A sequential calibration method based on pseudo-invariant regions is used, with F16 as the standard sensor, employing a univariate quadratic nonlinear regression equation. Where DN' is the result of mutual correction, DN0 is the original image gray value, and a, b, and c are the coefficients of the regression equation. The data of other sensors (F15, F14, F12, F10, and F18) are calculated in turn.

[0068] The DMSP / OLS has six sensors (F10, F12, F14, F15, F16, and F18); using F16 as the standard sensor, the specific calibration process for the other five sensors is as follows:

[0069] (1) Use the data of F16 to correct the data of F15. The years of overlap are 2004-2007. Use F16 2004 —F16 2007 data as the reference data set, and based on the least square method, use F15 2004 —F15 2007 data set to model a one-dimensional quadratic regression with the reference data set, and the regression equation is DN 16 = a x DN 2 15 + b x DN 15 + c, where DN 16 is the DN value of the F16 2004 —F16 2007 reference data set; DN 15 is the DN value of the F15 2004 —F15 2007 data set; a, b, and c are parameters determined in the fitting process.

[0070] (2) Use the parameters a, b, and c obtained to construct the equation DN' = a x DN 2 0 + b x DN0 + c for correcting the sensor F15, where DN0 is the DN value of F15 before correction, and DN' is the DN value after correction. Use this equation to process the F15 2000 —F15 2007 light data set, and obtain the mutually corrected data set F15'.

[0071] (3) Use the F15' data set to correct the data set of the sensor F14, and the method is the same as above, and obtain F14'. In this way, sequentially correct to obtain the data sets F12' and F10'.

[0072] (4) Correct the sensor F18. Since there is no overlap year data between F16 and F18, use the F16 2009 data to perform one-dimensional quadratic regression modeling with the F18 2010 data: DN 18 = a x DN 2 16 + b x DN 15 + c, where DN 18 is the DN value of the F18 2010 data, DN 18 is the DN value of the F18 2010 data, and a, b, and c are parameters determined in the fitting process. Then use the established equation to correct the data of each period of F18, and obtain the mutually corrected data set F18'.

[0073] (5) The fitting parameters a, b, and c of each period are shown in the following table:

[0074] Table 1 Parameters of sensor correction regression model

[0075]

[0076] S3: Data continuity correction on the DMSP / OLS night light data set after inter-sensor correction.

[0077] The F10-F18 annual light image data of DMSP / OLS in 1992-2013 is calibrated between different years, and the pixel-by-pixel averaging method is used The light image data of the sensor overlapping years (1994, 1997-2007) is processed to obtain 22 annual DMSP / OLS light image data of each year in 1992-2013, DN (y r, i) DN represents the pixel DN value of the i-th grid point of the night light data of a certain year, S1 and S2 represent two different sensors. The stable F16 light image data of 2009 is selected as the reference to calibrate the continuity of the 22 annual DMSP / OLS light image data. For the light image data before 2009, the first formula The DMSP / OLS light image data of 2008 to 1992 is calibrated year by year, and the pixel-by-pixel light brightness in the light image data of each year before 2009 should decrease year by year; for the data after 2009, the second formula The DMSP / OLS light image data of 2010 to 2013 is calibrated year by year, and the pixel-by-pixel light brightness in the light image data of each year after 2009 should increase year by year. Wherein, yr-1 takes the value range of 1992, 1993, …, 2008, DN (yr-1,i) DN (yr，i) DN is the pixel DN value of the i-th grid point of the night light data of the previous year; yr+1 takes the value range of 2010, 2011, 2012, 2013, DN (yr，i) DN is the pixel DN value of the i-th grid point of the night light data of the next year.

[0078] Among them, step S3 is summarized as:

[0079] (1) According to the DMSP / OLS night light data set after inter-sensor correction, the DMSP / OLS night light data set of the sensor overlapping years is processed by using the pixel-by-pixel averaging method to obtain the average processed DMSP / OLS night light data set.

[0080] (2) According to the DMSP / OLS night light data set corresponding to the reference sensor as the reference, the average processed DMSP / OLS night light data set is calibrated continuously.

[0081] According to the DMSP / OLS night light data set corresponding to the reference sensor as the reference, the average processed DMSP / OLS night light data set is calibrated continuously, specifically comprising:

[0082] 1) For the DMSP / OLS night light data set of the years before the year corresponding to the reference sensor, the first formula is used to calibrate continuously year by year.

[0083] 2) For the DMSP / OLS night light data set of the years after the year corresponding to the reference sensor, the second formula is used to calibrate continuously year by year.

[0084] S4: The pre-processed NPP / VIIRS night light data set is subjected to data noise reduction processing. Specifically:

[0085] The pre-processed NPP / VIIRS night light data set is screened by T0.3 mask method to obtain the noise-reduced NPP / VIIRS night light data set.

[0086] Taking the Guangdong-Hong Kong-Macao Greater Bay Area as an example, 9 NPP / VIIRS night light image data of 2013-2021 are screened by T0.3 mask method to obtain light pixels with light brightness above 0.3*10-9W·cm-2·sr-1 light DN value and below the maximum light DN value of Beijing, Shanghai or Guangzhou in the corresponding year. T0.3 mask is used to reduce the NPP / VIIRS data, and the influence of unstable light source and background noise is eliminated.

[0087] S5: The pre-processed NPP / VIIRS night light data set is subjected to continuous correction.

[0088] Taking the NPP / VIIRS night light image data of 2013 as the reference, the formula is used to calibrate the NPP / VIIRS night light image data of 2014 to 2021 year by year, and the light brightness in the light image data of each year after 2013 should be increased year by year.

[0089] S6: applying the inverse hyperbolic sine transformation to the data fitting model between the corrected DMSP / OLS night light data set and the corrected NPP / VIIRS night light data set in the overlapping period data set, and applying the data fitting model to the non-overlapping period NPP / VIIRS night light data set to obtain the non-overlapping period fitted DMSP / OLS night light data set; the overlapping period fitted DMSP / OLS night light data set and the non-overlapping period fitted DMSP / OLS night light data set constitute a complete reconstructed light data set.

[0090] Firstly, the DMSP / OLS and NPP / VIIRS light image data in 2013, the overlapping year of DMSP / OLS and NPP / VIIRS data, are selected; secondly, the 2013 DMSP / OLS light image data is fitted based on the 2013 NPP / VIIRS data, the inverse hyperbolic sine transformation is applied to the 2013 NPP / VIIRS data, NPP / VIIRS new is the DN value of the pixel brightness of the NPP / VIIRS data after the inverse hyperbolic sine transformation processing; and the sigmoid logic function (a, b, c are model fitting coefficients) is used to fit the 2013 DMSP / OLS data, an optimal fitting model is established, and the fitting model result is as shown in Figure 2 (b). Figure 2 (a) is the application of the logic function model to the fitting of the logarithmic transformed NPP / VIIRS data; Figure 2 (b) is the application of the logic function model to the fitting of the inverse hyperbolic sine (IHS) transformed NPP / VIIRS data; Figure 2 (c) is the application of the quadratic function model to the fitting of the DMSP / OLS data after the HSI method desaturation; Figure 2 (d) is the application of the quadratic function model to the fitting of the DMSP / OLS data after the VANUI method desaturation. Then, the 2014-2021 annual NPP / VIIRS light image data is input, and the fitting model is used for reconstruction, and the 2013-2021 reconstructed light data set is output. The fitting of the 2014-2021 NPP / VIIRS data is to use the model to fit and calculate the 2014-2021 annual NPP / VIIRS data, so as to obtain the reconstructed 2014-2021 series light data set.

[0091] Among them, the process of deriving the data fitting model in step S6 is summarized as:

[0092] (1) Determine the overlapping period dataset between the corrected DMSP / OLS night light dataset and the corrected NPP / VIIRS night light dataset.

[0093] (2) Perform the inverse hyperbolic sine transformation on the NPP / VIIRS night light dataset of the overlapping period.

[0094] (3) Fit the sigmoid logistic function to the inverse hyperbolic sine transformed NPP / VIIRS night light dataset of the overlapping period and the DMSP / OLS night light dataset of the overlapping period to obtain the data fitting model; the inverse hyperbolic sine transformed NPP / VIIRS night light dataset of the overlapping period is the fitted DMSP / OLS night light dataset of the overlapping period.

[0095] S7: Integrate the reconstructed light data set and the DMSP / OLS night light data set to obtain the multi-source integrated complete data, output the data of 2021 and visualize it to obtain Figure 3 , and output and visualize the data of other years to obtain similar images Figure 3 .

[0096] Integrate the reconstructed light data set from 2013 to 2021 with the DMSP / OLS light image data set from 1992 to 2013 to obtain the complete 1992-2021 Guangdong-Hong Kong-Macao Greater Bay Area light image data set.

[0097] The method is also applicable to the integration of long time series of DMSP / OLS and NPP / VIRS two-generation night light data in other areas outside the Guangdong-Hong Kong-Macao Greater Bay Area.

[0098] According to the above light data reconstruction method, the integration of long time series of DMSP / OLS and NPP / VIRS two-generation night light data can be realized.

[0099] Taking the Guangdong-Hong Kong-Macao Greater Bay Area as an example, in the embodiment, the DMSP / OLS and NPP / VIRS two kinds of night light data are based on, through the T0.3 mask noise reduction and the NPP / VIIRS after inverse hyperbolic sine transformation processing IHS(NPP / VIIRS) fitting DMSP / OLS, the optimal fitting model is obtained, and the model is used to fit the NPP / VIIRS data set to obtain the 2013-2021 series light data set, and then the DMSP / OLS light data set from 1992 to 2013 is integrated to obtain the Guangdong-Hong Kong-Macao Greater Bay Area 1992-2021 continuous night light data set. Provide a scientific basis for constructing a scientific and reasonable urban system structure and optimizing the spatial pattern of urban expansion. And the verification result shows that the correlation R2 of the light index of the method with GDP is 0.96, and the correlation R2 of the light index of the method with the urban built-up area is 0.93. Therefore, the method has the characteristics of high precision and good fitting effect, and better solves the problems existing in the prior art, and effectively improves the accuracy of the long-time series light data integration reconstruction method.

[0100] Compared with the prior art, the present application has the following advantages:

[0101] (1) The present application obtains the optimal fitting model by fitting DMSP / OLS with T0.3 mask noise reduction and NPP / VIIRS after inverse hyperbolic sine transformation processing (IHS(NPP / VIIRS)), and the fitting effect of the model is better than that of the same type method.

[0102] (2) The present application can solve the fault problem between the two generations of light data DMSP / OLS and NPP / VIRS, and obtain a set of Guangdong-Hong Kong-Macao Greater Bay Area 1992-2021 continuous night light data set product, which has high correlation with GDP, urban built-up area and other statistical data.

[0103] Example 2

[0104] The embodiment provides a multi-source long-time series night light data reconstruction system, which comprises:

[0105] A data acquisition and preprocessing module is configured to acquire DMSP / OLS night light data set and NPP / VIIRS night light data set of a target region and perform spatial resolution unification preprocessing.

[0106] A first correction module is configured to correct the preprocessed DMSP / OLS night light data set.

[0107] A second correction module is configured to correct the DMSP / OLS night light data set after sensor correction.

[0108] a third correction module configured to perform continuity correction on the pre-processed NPP / VIIRS night light dataset.

[0109] a fitting module configured to perform data fitting by applying an inverse hyperbolic sine transformation on the overlapping period dataset between the corrected DMSP / OLS night light dataset and the corrected NPP / VIIRS night light dataset, to obtain a data fitting model.

[0110] a data reconstruction module configured to apply the data fitting model to the non-overlapping period NPP / VIIRS night light dataset to obtain a non-overlapping period fitted DMSP / OLS night light dataset; the overlapping period fitted DMSP / OLS night light dataset and the non-overlapping period fitted DMSP / OLS night light dataset constitute a complete reconstructed light dataset.

[0111] a multi-source data integration module configured to integrate the reconstructed light dataset and the DMSP / OLS night light dataset to obtain a multi-source integrated complete dataset.

[0112] The same or similar parts among various embodiments can be referred to each other. For the system disclosed by the embodiments, since it corresponds to the method disclosed by the embodiments, the description is relatively simple, and the relevant part can be referred to the method part.

[0113] The principles and implementation manners of the present application are described by using specific examples in the present application, and the above embodiment description is only used to help understand the method of the present application and its core idea; meanwhile, for the person skilled in the art, according to the idea of the present application, the specific implementation manner and application range can be changed. In conclusion, the content of the present application should not be understood as the limitation of the present application.

Claims

1. A method for reconstructing multi-source long-term nighttime light data, characterized in that, The method includes: Acquire the DMSP / OLS nighttime light dataset and NPP / VIIRS nighttime light dataset for the target area and perform spatial resolution uniform preprocessing; Perform cross-calibration of data between various sensors on the preprocessed DMSP / OLS nighttime light dataset; Perform data continuity correction on the DMSP / OLS nighttime light dataset after inter-sensor calibration; Continuity correction was performed on the preprocessed NPP / VIIRS nighttime light dataset; The inverse hyperbolic sine transform was applied to fit the overlapping time period datasets between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset to obtain the data fitting model. The data fitting model is applied to the NPP / VIIRS nighttime light datasets for non-overlapping periods to obtain the DMSP / OLS nighttime light datasets fitted for non-overlapping periods; the DMSP / OLS nighttime light datasets fitted for overlapping periods and the DMSP / OLS nighttime light datasets fitted for non-overlapping periods constitute a complete reconstructed light dataset. The reconstructed light dataset and the DMSP / OLS nighttime light dataset are integrated to obtain complete data after multi-source integration; Specifically, an inverse hyperbolic sine transform was applied to fit the overlapping time periods between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset to obtain a data fitting model, which includes: Identify the overlapping time periods between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset; Inverse hyperbolic sine transform is performed on the NPP / VIIRS nighttime light dataset during overlapping time periods; The sigmoid logistic function is used to fit the NPP / VIIRS nighttime light dataset after inverse hyperbolic sine transform of the overlapping period and the DMSP / OLS nighttime light dataset of the overlapping period to obtain the data fitting model; the NPP / VIIRS nighttime light dataset after inverse hyperbolic sine transform of the overlapping period is the DMSP / OLS nighttime light dataset fitted by the overlapping period. The expression for the inverse hyperbolic sine transform is: The expression for the sigmoid logical function is: NPP / VIIRS new DN values ​​are the pixel brightness values ​​after NPP / VIIRS data have undergone inverse hyperbolic sine transform processing; a, b, and c are model fitting coefficients.

2. The method according to claim 1, characterized in that, The DMSP / OLS nighttime light dataset and NPP / VIIRS nighttime light dataset for the target region were acquired and preprocessed to achieve spatial resolution uniformity. Specifically, this included: The DMSP / OLS nighttime light dataset for the first preset year and the NPP / VIIRS nighttime light dataset for the second preset year are respectively reprojected onto the Albers equal-area projection of the preset region; Pixel resampling was performed on the reprojected DMSP / OLS nighttime light dataset and the reprojected NPP / VIIRS nighttime light dataset based on the nearest neighbor method. The resampled DMSP / OLS nighttime light dataset and the resampled NPP / VIIRS nighttime light dataset are respectively cropped according to the boundary data of the target area to obtain the preprocessed DMSP / OLS nighttime light dataset and the preprocessed NPP / VIIRS nighttime light dataset.

3. The method according to claim 1, characterized in that, The preprocessed DMSP / OLS nighttime light dataset undergoes cross-calibration between various sensors, specifically including: Based on the sequential calibration method of pseudo-invariant regions, the data of each sensor in each period corresponding to the preprocessed DMSP / OLS nighttime light dataset are calibrated to obtain the DMSP / OLS nighttime light dataset after inter-sensor correction.

4. The method according to claim 3, characterized in that, Based on the sequential calibration method of pseudo-invariant regions, the preprocessed DMSP / OLS nighttime light dataset is calibrated for each sensor and each period, specifically including: A reference sensor is determined from each sensor corresponding to the preprocessed DMSP / OLS nighttime light dataset; The sensor data of the non-benchmark sensor in each period are calculated sequentially using a univariate quadratic nonlinear regression equation to obtain the DMSP / OLS nighttime light dataset after inter-sensor correction.

5. The method according to claim 4, characterized in that, Data continuity correction is performed on the DMSP / OLS nighttime light dataset after inter-sensor calibration, specifically including: Based on the DMSP / OLS nighttime light dataset calibrated between sensors, the pixel-by-pixel averaging method is used to process the DMSP / OLS nighttime light dataset of the overlapping years of the sensors, resulting in the averaged DMSP / OLS nighttime light dataset. Based on the DMSP / OLS nighttime light dataset corresponding to the benchmark sensor, the averaged DMSP / OLS nighttime light dataset is continuously calibrated.

6. The method according to claim 5, characterized in that, Based on the DMSP / OLS nighttime light dataset corresponding to the benchmark sensor, continuous calibration is performed on the averaged DMSP / OLS nighttime light dataset, specifically including: For the DMSP / OLS nighttime light datasets from years preceding the year corresponding to the benchmark sensor, continuous calibration is performed year by year using the first formula; the first formula is: For the DMSP / OLS nighttime light datasets of the years following the year corresponding to the reference sensor, continuous calibration is performed year by year using the second formula; the second formula is: Where DN represents the DN value of a pixel; yr represents the year variable; yr-1 represents the year before yr; yr+1 represents the year after yr; and i represents the i-th grid point of the nighttime light data.

7. The method according to claim 6, characterized in that, Before performing continuity correction on the preprocessed NPP / VIIRS nighttime light dataset, the following steps are also included: The T0.3 mask method is used to filter out light pixels with light brightness above a first preset value and below a second preset value from the preprocessed NPP / VIIRS night light dataset to obtain the noise-reduced NPP / VIIRS night light dataset.

8. A system for reconstructing multi-source long-term nighttime light data, characterized in that, The system includes: The data acquisition and preprocessing module is used to acquire the DMSP / OLS nighttime light dataset and NPP / VIIRS nighttime light dataset of the target area and perform spatial resolution uniform preprocessing. The first calibration module is used to perform data mutual calibration between various sensors on the preprocessed DMSP / OLS nighttime light dataset. The second calibration module is used to perform data continuity calibration on the DMSP / OLS nighttime light dataset after inter-sensor calibration. The third correction module is used to perform continuity correction on the preprocessed NPP / VIIRS nighttime light dataset; The fitting module is used to apply an inverse hyperbolic sine transform to fit the data based on the overlapping time period dataset between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset, and obtain the data fitting model. Specifically, an inverse hyperbolic sine transform was applied to fit the overlapping time periods between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset to obtain a data fitting model, which includes: Identify the overlapping time periods between the corrected DMSP / OLS nighttime light dataset and the corrected NPP / VIIRS nighttime light dataset; Inverse hyperbolic sine transform is performed on the NPP / VIIRS nighttime light dataset during overlapping time periods; The sigmoid logistic function is used to fit the NPP / VIIRS nighttime light dataset after inverse hyperbolic sine transform of the overlapping period and the DMSP / OLS nighttime light dataset of the overlapping period to obtain the data fitting model; the NPP / VIIRS nighttime light dataset after inverse hyperbolic sine transform of the overlapping period is the DMSP / OLS nighttime light dataset fitted by the overlapping period. The expression for the inverse hyperbolic sine transform is: The expression for the sigmoid logical function is: NPP / VIIRS new These are the pixel brightness DN values ​​after NPP / VIIRS data have undergone inverse hyperbolic sine transform processing; a, b, and c are model fitting coefficients; The data reconstruction module is used to apply the data fitting model to the NPP / VIIRS nighttime light datasets of non-overlapping periods to obtain the DMSP / OLS nighttime light datasets fitted for non-overlapping periods; the DMSP / OLS nighttime light datasets fitted for overlapping periods and the DMSP / OLS nighttime light datasets fitted for non-overlapping periods constitute a complete reconstructed light dataset. The multi-source data integration module is used to integrate the reconstructed light dataset and the DMSP / OLS nighttime light dataset to obtain complete data after multi-source integration.

Images