Method and apparatus for producing monthly vegetation index products from the environment satellite-2

Abstract

This invention discloses a method and apparatus for producing monthly vegetation index products from the HJ-2 satellite, belonging to the field of remote sensing monitoring technology. The method includes: acquiring and preprocessing HJ-2 remote sensing data; calculating the vegetation index of each period of HJ-2 remote sensing data to obtain vegetation index images; synthesizing the maximum values ​​of all vegetation index images within each natural ten-day period to obtain a maximum value image; acquiring the maximum value images of the target month, the preceding month, and the following month, and sorting them chronologically; using the Savitzky-Golay filtering algorithm, seamlessly reconstructing the full-time-series vegetation index by sliding through a window on the sorted continuous time-series data; extracting the maximum value images of the three ten-day periods corresponding to the target month from the reconstructed continuous time-series data, and synthesizing the maximum values ​​to obtain the vegetation index product for the target month. This invention can generate high-precision, spatiotemporally seamless, and phenologically accurate HJ-2 monthly vegetation index products, comprehensively enhancing the product application value of HJ-2 satellite data.

Description

Technical Field

[0001] This invention relates to the field of remote sensing monitoring technology, specifically to a method and apparatus for producing monthly vegetation index products from the Environment-2 satellite. Background Technology

[0002] Vegetation indices are core remote sensing parameters characterizing changes in land surface vegetation cover, growth status, and biomass. They are widely used in fields such as ecological environment monitoring, ecological quality assessment, and evaluation of the effectiveness of ecological protection and restoration. The HJ-2 environmental remote sensing satellite is a high-resolution environmental remote sensing satellite with advantages such as high spatial resolution, high observation frequency, and wide coverage. It can achieve refined dynamic monitoring of regional vegetation and is an important data source for vegetation remote sensing.

[0003] However, the existing HJ-2 satellite vegetation index product production technology has significant flaws:

[0004] 1. HJ-2 satellite remote sensing images are easily affected by meteorological factors such as clouds, aerosols, and snow, resulting in a large number of missing pixels and noisy pixels in a single image, causing the time series data to be broken and incomplete.

[0005] 2. Traditional monthly vegetation index products often use single-month data to directly synthesize or perform simple time-series interpolation, failing to fully utilize cross-month time-series correlation information, and are unable to accurately repair missing areas of monthly data, resulting in poor product spatial integrity. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method and apparatus for producing monthly vegetation index products from the HJ-2 satellite, which can generate high-precision, spatiotemporally seamless, and phenologically accurate HJ-2 monthly vegetation index products, thereby comprehensively enhancing the product application value of HJ-2 satellite data.

[0007] The technical solution provided by this invention is as follows:

[0008] A method for producing a monthly vegetation index product from the Environment-2 satellite, the method comprising:

[0009] S1: Acquire remote sensing data from the Environment-2 satellite and perform preprocessing;

[0010] S2: Calculate the vegetation index of each period of the Environment-2 satellite remote sensing data to obtain vegetation index images;

[0011] S3: Using the natural ten-day period as the basic time unit, the maximum value of all valid vegetation index images within each ten-day period is synthesized to obtain the maximum value image;

[0012] S4: For the target month to be produced, obtain the maximum value images of the target month and the months preceding and following the target month, and sort them in chronological order to construct continuous time series data;

[0013] S5: Using the Savitzky-Golay filtering algorithm, the continuous time series data is traversed window by window to achieve seamless reconstruction of the full time series vegetation index;

[0014] S6: Extract the maximum image of the three ten-day periods corresponding to the target month from the reconstructed continuous time series data, and synthesize the maximum value to obtain the vegetation index product of the target month.

[0015] Furthermore, S5 includes:

[0016] S51: Calculate the temporal standard deviation of each pixel position based on the continuous time series data, and classify pixels with a temporal standard deviation greater than a set threshold as high noise regions, and other pixels as low noise regions.

[0017] S52: Set a low-order polynomial order for the high-noise region and a high-order polynomial order for the low-noise region.

[0018] S53: Construct the least squares fitting equation according to the Savitzky-Golay filtering algorithm, fit the temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill the missing pixels to obtain the temporal reconstruction result of the current round.

[0019] The least squares fitting equation for the high-noise region uses a low-order polynomial, while the least squares fitting equation for the low-noise region uses a high-order polynomial.

[0020] S54: Calculate the residual between the continuous time series data and the time series reconstruction result of the current round, use the 3σ criterion to identify residual noise abnormal observation points, and re-mark the abnormal observation points as missing pixels; return to S53 and repeat the execution until the set iteration termination condition is reached.

[0021] Furthermore, S53 also includes:

[0022] The upper envelope constraint rule is adopted to ensure that the temporal reconstruction result of the current round does not have negative values ​​or abnormally high values ​​with abnormal physical meaning.

[0023] Furthermore, the lower-order polynomial has an order of 2, and the higher-order polynomial has an order of 3 to 5.

[0024] Furthermore, the preprocessing includes:

[0025] Sensor system errors were eliminated through radiometric calibration; atmospheric correction was performed using a 6S model to remove atmospheric scattering, absorption, and aerosol interference; and geometric correction was performed through ground control point matching and a quadratic polynomial model.

[0026] Furthermore, the method also includes:

[0027] S7: Execute S4~S6 sequentially according to the natural month to obtain the vegetation index products for each month.

[0028] Furthermore, the method also includes:

[0029] S8: For vegetation index products for each month, abnormal pixels are removed by combining regional vegetation phenology thresholds, and accuracy is verified based on ground-measured sample points.

[0030] A device for producing monthly vegetation index products from the Environment-2 satellite, the device comprising:

[0031] The data acquisition module is used to acquire and preprocess remote sensing data from the Environment-2 satellite.

[0032] The vegetation index calculation module is used to calculate the vegetation index of each period of the Environment-2 satellite remote sensing data to obtain vegetation index images;

[0033] The maximum value synthesis module is used to synthesize the maximum value of all valid vegetation index images within each ten-day period, based on the natural ten-day period as the basic time unit, to obtain the maximum value image.

[0034] The 90-month cross-month spatiotemporal reconstruction module is used to acquire the maximum value images of the target month, the month before the target month, and the month after the target month for the target month to be produced, and sort them in chronological order to construct continuous time series data;

[0035] The sliding filter module is used to seamlessly reconstruct the full-time vegetation index by sliding through the continuous time-series data window by window using the Savitzky-Golay filtering algorithm.

[0036] The monthly maximum value synthesis module is used to extract the three-ten-day maximum value image corresponding to the target month from the reconstructed continuous time series data, and perform maximum value synthesis to obtain the vegetation index product of the target month.

[0037] Furthermore, the sliding filter module includes:

[0038] The region division unit is used to calculate the temporal standard deviation of each pixel position based on the continuous time series data, and to divide pixels with a temporal standard deviation greater than a set threshold into high noise regions, and other pixels into low noise regions.

[0039] A polynomial order setting unit is used to set a low-order polynomial order for the high-noise region and a high-order polynomial order for the low-noise region.

[0040] The temporal smoothing reconstruction unit is used to construct the least squares fitting equation according to the Savitzky-Golay filtering algorithm, fit the temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill the missing pixels to obtain the temporal reconstruction result of the current round.

[0041] The least squares fitting equation for the high-noise region uses a low-order polynomial, while the least squares fitting equation for the low-noise region uses a high-order polynomial.

[0042] The iterative unit is used to calculate the residual between the continuous time series data and the time series reconstruction result of the current round, and uses the 3σ criterion to identify residual noise abnormal observation points and re-mark the abnormal observation points as missing pixels; it returns to the time series smoothing reconstruction unit to repeat the process until the set iteration termination condition is met.

[0043] Furthermore, the timing smoothing reconstruction unit is also used for:

[0044] The upper envelope constraint rule is adopted to ensure that the temporal reconstruction result of the current round does not have negative values ​​or abnormally high values ​​with abnormal physical meaning.

[0045] Furthermore, the lower-order polynomial has an order of 2, and the higher-order polynomial has an order of 3 to 5.

[0046] Furthermore, the preprocessing includes:

[0047] Sensor system errors were eliminated through radiometric calibration; atmospheric correction was performed using a 6S model to remove atmospheric scattering, absorption, and aerosol interference; and geometric correction was performed through ground control point matching and a quadratic polynomial model.

[0048] Furthermore, the device also includes:

[0049] The monthly production module is used to sequentially execute the ninety-monthly spatiotemporal reconstruction module, the sliding filter module, and the monthly maximum value synthesis module according to the natural month sequence to obtain the vegetation index products for each month.

[0050] Furthermore, the device also includes:

[0051] The quality control module is used to remove abnormal pixels from the vegetation index products for each month by combining regional vegetation phenology thresholds, and to verify the accuracy based on ground-based measured sample points.

[0052] The present invention has the following beneficial effects:

[0053] This invention aims to overcome the shortcomings of existing technologies in HJ-2 satellite vegetation index, such as discontinuous temporal series, severe data gaps, distorted phenological characteristics, and low standardization of monthly products. It constructs a three-period, nine-day-a-week time-series reconstruction architecture, adapted to the characteristics of HJ-2 satellite data, and specifically addresses the industry pain points of cloud pollution, temporal breaks, and severe data gaps in HJ-2 high-resolution remote sensing satellite vegetation data. It fills the gap in the technology for monthly standardized production of HJ-2 satellite monthly vegetation indices. This invention employs a multi-scale temporal series optimization strategy using the maximum value synthesis processing of ten-day periods and nine-day-a-week cross-month reconstruction. Compared to traditional single-month interpolation methods, it fully utilizes the phenological correlation information between preceding and following months, significantly improving the accuracy of missing data reconstruction. The spatial integrity of the product can reach 100%, and the temporal continuity is significantly improved. The resulting vegetation index product has high spatial resolution, good temporal consistency, and accurate phenological characteristics, and can be widely applied in various fields such as regional ecological quality assessment, vegetation dynamic evolution analysis, crop growth and yield prediction, and desertification monitoring.

[0054] This invention utilizes a technical system of combining the maximum value of a ten-day period, reconstructing the spatiotemporal data across nine ten-day periods, implementing Savitzky-Golay sliding iterative optimization, and accurately combining the maximum value of a month to ultimately generate a high-precision, spatiotemporally seamless, and phenologically accurate HJ-2 monthly vegetation index product, thereby comprehensively enhancing the product application value of HJ-2 satellite data. Attached Figure Description

[0055] Figure 1 This is a flowchart of the method for producing monthly vegetation index products from the Environment-2 satellite according to the present invention.

[0056] Figure 2 This is a schematic diagram of the production device for the monthly vegetation index product of the Environment-2 satellite according to the present invention. Detailed Implementation

[0057] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0058] This invention provides a method for producing monthly vegetation index products from the HJ-2 satellite, which is a method for producing monthly vegetation indexes based on three-period ninety-day time series reconstruction. Figure 1 As shown, the method includes:

[0059] S1: Acquire remote sensing data from the Environment-2 satellite and perform preprocessing.

[0060] Traditional monthly vegetation index products often use direct synthesis of single-month data or simple time-series interpolation. They cannot repair missing data areas within a single month (especially at the beginning and end of the month), resulting in poor spatial integrity of the product.

[0061] This invention employs a processing method of "pre-month + target month + subsequent month," thus resolving the issue of missing data within a month. Therefore, this invention should acquire remote sensing data from the Environment-2 satellite for at least three months: the pre-month, the target month, and the subsequent month.

[0062] For the production of single-month vegetation index products, only data from the preceding month, target month, and subsequent month need to be acquired. For continuous monthly production, data from each month needs to be acquired. In this invention, it is not necessary to wait for all the environmental data from the Environment-2 satellite to be available before production; data can be acquired and produced month by month. For example, if the latest monthly data is for June, then vegetation index products for May can be produced.

[0063] After acquiring remote sensing data from the Environment-2 satellite for each period, radiometric calibration was performed sequentially to eliminate sensor system errors. A 6S model was used to implement precise atmospheric correction, removing atmospheric scattering, absorption, and aerosol interference to obtain accurate surface reflectance information. Geometric fine correction was completed through ground control point matching and a quadratic polynomial model to ensure the spatial registration accuracy of multi-period images, laying a precise data foundation for subsequent vegetation index calculations.

[0064] S2: Calculate the vegetation index of each scene of the Environment 2 satellite remote sensing data to obtain the vegetation index image.

[0065] The vegetation index mentioned in this invention may be the Normalized Difference Vegetation Index (NDVI), the Enhanced Vegetation Index (EVI), or other vegetation indices.

[0066] S3: Using natural ten-day periods as the basic time unit, the maximum value of all valid vegetation index images within each ten-day period is synthesized to obtain the maximum value image.

[0067] This step is used for maximum value synthesis and noise reduction within a ten-day period. Specifically, it compares the values ​​of all valid vegetation index images within each ten-day period at the same pixel location, and uses the maximum value as the value at that pixel location to achieve maximum value synthesis. This process removes low-value noise pixels caused by cloud pollution, snow and ice blockage, and sensor anomalies, generating a time-ordered and noise-removed ten-day vegetation index base dataset (i.e., maximum value image), thus achieving preliminary optimization of short-term observation errors.

[0068] S4: For the target month to be produced, obtain the maximum value images of the target month and the months preceding and following the target month, and sort them in chronological order to construct continuous time series data.

[0069] This step is used to construct a three-period, ninety-day time series window: For the target monthly vegetation index product to be produced, it breaks through the limitations of single-month data and constructs a three-period, ninety-day continuous time series data window consisting of "three ten-day periods of the preceding month + three ten-day periods of the target month + three ten-day periods of the following month" using the preceding month, the target month, and the following month as the three core time series periods. This fully utilizes the continuity of vegetation time series and the correlation with phenology to provide sufficient time series support for the reconstruction of missing data time series and solve the problems of insufficient single-month data and low reconstruction accuracy.

[0070] S5: Using the Savitzky-Golay filtering algorithm, continuous time-series data are traversed window by window to achieve seamless reconstruction of the full-time vegetation index.

[0071] This invention employs the Savitzky-Golay filtering algorithm to achieve seamless reconstruction of the full-time vegetation index by sliding through a window-by-window process on continuous time-series data of ninety-year periods. Combined with time-series weights, it preserves the dynamic changes in vegetation phenology and accurately repairs missing data areas, resulting in reconstructed vegetation index data that is spatiotemporally continuous across the entire domain.

[0072] S6: Extract the maximum image of the three ten-day periods corresponding to the target month from the reconstructed continuous time series data, and synthesize the maximum value to obtain the vegetation index product of the target month.

[0073] Specifically, vegetation index data corresponding to the target month can be accurately extracted from the reconstructed ninety-day continuous data, and the maximum value synthesis process can be performed again to retain the observation information of the optimal vegetation growth status within the month, further suppress residual noise, and generate a high-precision, full-coverage vegetation index product for the target month.

[0074] This invention aims to overcome the shortcomings of existing technologies in HJ-2 satellite vegetation index, such as discontinuous temporal series, severe data gaps, distorted phenological characteristics, and low standardization of monthly products. It constructs a three-period, nine-day-a-week time-series reconstruction architecture, adapted to the characteristics of HJ-2 satellite data, and specifically addresses the industry pain points of cloud pollution, temporal breaks, and severe data gaps in HJ-2 high-resolution remote sensing satellite vegetation data. It fills the gap in the technology for monthly standardized production of HJ-2 satellite monthly vegetation indices. This invention employs a multi-scale temporal series optimization strategy using the maximum value synthesis processing of ten-day periods and nine-day-a-week cross-month reconstruction. Compared to traditional single-month interpolation methods, it fully utilizes the phenological correlation information between preceding and following months, significantly improving the accuracy of missing data reconstruction. The spatial integrity of the product can reach 100%, and the temporal continuity is significantly improved. The resulting vegetation index product has high spatial resolution, good temporal consistency, and accurate phenological characteristics, and can be widely applied in various fields such as regional ecological quality assessment, vegetation dynamic evolution analysis, crop growth and yield prediction, and desertification monitoring.

[0075] In summary, this invention, through a technical system of combining and processing the maximum value of a ten-day period, reconstructing the spatiotemporal data across nine ten-day periods, implementing Savitzky-Golay sliding iterative optimization, and accurately combining the maximum value of a month, ultimately generates a high-precision, spatiotemporally seamless, and phenologically accurate HJ-2 monthly vegetation index product, comprehensively enhancing the product application value of HJ-2 satellite data.

[0076] Existing conventional time-series filtering algorithms have fixed windows and weak spatiotemporal adaptability, easily leading to over-smoothing of vegetation phenological details and distortion of key features such as vegetation growth cycles and peak growth. To address this issue, this invention employs an improved adaptive variable-order Savitzky-Golay filtering algorithm, the specific implementation process of which is as follows:

[0077] S51: Calculate the temporal standard deviation of each pixel position based on continuous time series data, and classify pixels with a temporal standard deviation greater than a set threshold as high noise regions, and other pixels as low noise regions.

[0078] S52: Sets the polynomial order to a low degree for high-noise regions and a polynomial order to a high degree for low-noise regions.

[0079] This invention quantifies noise levels such as cloud cover and aerosols by statistically analyzing the time series standard deviation within a single pixel over a ninety-day period. For cloudy areas with few observations (high-noise regions), a second-order low-order polynomial fitting is adaptively selected to avoid overfitting distortion caused by high-order fitting. For clear-sky, lush vegetation areas (low-noise regions), a third- to fifth-order high-order fitting is used to reconstruct detailed phenological dynamics. This ensures accurate repair of missing data areas while avoiding excessive smoothing that could lead to the loss of phenological details.

[0080] S53: Construct the least squares fitting equation according to the Savitzky-Golay filtering algorithm, fit the temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill the missing pixels to obtain the temporal reconstruction result of the current round.

[0081] Specifically, following the standard Savitzky-Golay convolution coefficient solution method and combining the adaptively selected optimal polynomial order, a least squares fitting equation is constructed within a ninety-year time window to solve the time-series smoothing curve. Among them, the least squares fitting equation in the high-noise region adopts the set low-order polynomial order, while the least squares fitting equation in the low-noise region adopts the set high-order polynomial order.

[0082] Then, the missing pixels for clouds, snow, and shadows are filled using the fitted curves through interpolation to obtain the temporal reconstruction results for the current round. An upper envelope constraint rule is employed to ensure that the vegetation index temporal series does not exhibit physically anomalous negative or abnormally high values.

[0083] S54: Calculate the residual between the continuous time series data and the time series reconstruction result of the current round, use the 3σ criterion to identify residual noise abnormal observation points, and re-mark the abnormal observation points as missing pixels; return to S53 and repeat the execution until the set iteration termination condition is reached.

[0084] This step utilizes a residual iterative correction strategy to optimize the time series fitting results. Specifically: based on an iterative denoising mechanism, the residual between the continuous time series data and the time series reconstruction result of the current round is calculated. The 3σ criterion is used to identify residual noise anomalous observation points, and outliers are re-labeled as missing pixels. The Savitzky-Golay fitting iteration is then executed again based on an adaptive optimal order. The iteration termination condition is set: the mean residual is below 0.01 or the maximum number of iterations is reached, and the final smooth and continuous ninety-year time series sequence is output.

[0085] This invention employs an adaptive variable-order Savitzky-Golay filtering algorithm, which balances noise smoothing with phenological feature preservation. It effectively avoids the distortion of vegetation growth peaks and phenological abrupt change characteristics caused by excessive smoothing in traditional filtering algorithms, resulting in higher authenticity and reliability of product data.

[0086] Existing technologies lack a standardized ten-day to monthly progressive reconstruction and sliding iteration production system, making it difficult to achieve automated and standardized high spatiotemporal resolution vegetation index product output on a monthly basis, and thus failing to meet the application needs of long-term, refined ecological monitoring.

[0087] To solve this problem, the method of the present invention further includes:

[0088] S7: Execute S4~S6 sequentially according to the natural month to obtain the vegetation index products for each month.

[0089] This invention enables monthly sliding iterative batch production. Specifically, it updates the ninety-year time series window sequentially according to the natural month sequence, automatically iterates and executes the window construction, data reconstruction, and monthly synthesis process, and completes the production of the whole-domain vegetation index product month by month, forming a standardized, long-time series, spatiotemporally continuous 16m resolution monthly vegetation index product sequence, thus realizing automated batch production.

[0090] This invention constructs an automated production system that slides and iterates monthly. The processing flow is standardized and reusable, and it can quickly generate long-term monthly vegetation index products in batches. The production efficiency is much higher than the traditional manual periodic processing method, and it can adapt to the production needs of large-scale and normalized ecological remote sensing monitoring products.

[0091] As an improvement to an embodiment of the present invention, the method further includes:

[0092] S8: For vegetation index products for each month, abnormal pixels are removed by combining regional vegetation phenology thresholds, and accuracy is verified based on ground-measured sample points.

[0093] This step is used for standardized product quality control output: abnormal pixels are removed by combining regional vegetation phenology thresholds, and accuracy verification is completed based on ground-measured sample points. After meeting quality control indicators such as effective pixel coverage, coefficient of determination, and root mean square error, monthly vegetation index products with 16m spatial resolution in standardized GeoTIFF format under the CGCS2000 coordinate system are output, and product naming rules are unified to achieve standardized product archiving and sharing.

[0094] Finally, the method of this invention can be embedded into a localized parameter inversion process to carry out stable inversion of key ecological parameters such as vegetation cover index (FVC) based on continuous monthly vegetation index sequences.

[0095] This invention also provides a device for producing monthly vegetation index products from the Environment-2 satellite, such as... Figure 2 As shown, the device includes:

[0096] Data acquisition module 1 is used to acquire remote sensing data from the Environment-2 satellite and perform preprocessing.

[0097] The vegetation index calculation module 2 is used to calculate the vegetation index of each period of the Environment-2 satellite remote sensing data to obtain vegetation index images.

[0098] The maximum value synthesis module 3 is used to synthesize the maximum value of all valid vegetation index images within each ten-day period, based on the natural ten-day period as the basic time unit, to obtain the maximum value image.

[0099] The 90-month spatiotemporal reconstruction module 4 is used to acquire the maximum value images of the target month, the month before the target month, and the month after the target month for the target month to be produced, and sort them in chronological order to construct continuous time series data.

[0100] The sliding filter module 5 is used to seamlessly reconstruct the full-time vegetation index by sliding through the window-by-window traversal of continuous time-series data using the Savitzky-Golay filtering algorithm.

[0101] The monthly maximum value synthesis module 6 is used to extract the three-ten-day maximum value image corresponding to the target month from the reconstructed continuous time series data, and perform maximum value synthesis to obtain the vegetation index product of the target month.

[0102] As one implementation, the sliding filter module includes:

[0103] The region division unit is used to calculate the temporal standard deviation of each pixel location based on continuous time-series data, and to classify pixels with a temporal standard deviation greater than a set threshold as high-noise regions, and other pixels as low-noise regions.

[0104] The polynomial order setting unit is used to set a low-order polynomial order for high-noise regions and a high-order polynomial order for low-noise regions.

[0105] Among them, the order of the low-order polynomial is 2, and the order of the high-order polynomial is 3 to 5.

[0106] The temporal smoothing reconstruction unit is used to construct a least-squares fitting equation according to the Savitzky-Golay filtering algorithm, fit a temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill missing pixels to obtain the temporal reconstruction result of the current round. An upper envelope constraint rule is also used to ensure that the temporal reconstruction result of the current round does not contain physically anomalous negative values ​​or abnormally high values.

[0107] The least squares fitting equation for the high-noise region uses a low-order polynomial, while the least squares fitting equation for the low-noise region uses a high-order polynomial.

[0108] The iterative unit is used to calculate the residual between the continuous time series data and the time series reconstruction result of the current round. It uses the 3σ criterion to identify residual noise abnormal observation points and re-marks the abnormal observation points as missing pixels. It returns to the time series smoothing reconstruction unit to repeat the process until the set iteration termination condition is met.

[0109] Specifically, the preprocessing includes:

[0110] Sensor system errors were eliminated through radiometric calibration; atmospheric correction was performed using a 6S model to remove atmospheric scattering, absorption, and aerosol interference; and geometric correction was performed through ground control point matching and a quadratic polynomial model.

[0111] As an improvement to an embodiment of the present invention, the apparatus further includes:

[0112] The monthly production module is used to sequentially execute the ninety-monthly spatiotemporal reconstruction module, the sliding filter module, and the monthly maximum value synthesis module according to the natural month sequence to obtain the vegetation index products for each month.

[0113] The quality control module is used to remove abnormal pixels from the vegetation index products for each month by combining regional vegetation phenology thresholds, and to verify the accuracy based on ground-based measured sample points.

[0114] The apparatus provided in this embodiment of the invention operates on the same principle and produces the same technical effects as the aforementioned method embodiments. For the sake of brevity, any parts not mentioned in the apparatus embodiments can be referred to the corresponding content in the aforementioned method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the apparatus and units described above can all be referred to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0115] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit them. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention.

Claims

1. A method for producing monthly vegetation index products from the Environment-2 satellite, characterized in that, The method includes: S1: Acquire remote sensing data from the Environment-2 satellite and perform preprocessing; S2: Calculate the vegetation index of each period of the Environment-2 satellite remote sensing data to obtain vegetation index images; S3: Using the natural ten-day period as the basic time unit, the maximum value of all valid vegetation index images within each ten-day period is synthesized to obtain the maximum value image; S4: For the target month to be produced, obtain the maximum value images of the target month and the months preceding and following the target month, and sort them in chronological order to construct continuous time series data; S5: Using the Savitzky-Golay filtering algorithm, the continuous time series data is traversed window by window to achieve seamless reconstruction of the full time series vegetation index; S6: Extract the maximum image of the three ten-day periods corresponding to the target month from the reconstructed continuous time series data, and synthesize the maximum value to obtain the vegetation index product of the target month.

2. The method for producing monthly vegetation index products from the Environment-2 satellite according to claim 1, characterized in that, S5 includes: S51: Calculate the temporal standard deviation of each pixel position based on the continuous time series data, and classify pixels with a temporal standard deviation greater than a set threshold as high noise regions, and other pixels as low noise regions. S52: Set a low-order polynomial order for the high-noise region and a high-order polynomial order for the low-noise region. S53: Construct the least squares fitting equation according to the Savitzky-Golay filtering algorithm, fit the temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill the missing pixels to obtain the temporal reconstruction result of the current round. The least squares fitting equation for the high-noise region uses a low-order polynomial, while the least squares fitting equation for the low-noise region uses a high-order polynomial. S54: Calculate the residual between the continuous time series data and the time series reconstruction result of the current round, use the 3σ criterion to identify residual noise abnormal observation points, and re-mark the abnormal observation points as missing pixels; return to S53 and repeat the execution until the set iteration termination condition is reached.

3. The method for producing monthly vegetation index products from the Environment-2 satellite according to claim 2, characterized in that, S53 further includes: The upper envelope constraint rule is adopted to ensure that the temporal reconstruction result of the current round does not have negative values ​​or abnormally high values ​​with abnormal physical meaning.

4. The method for producing monthly vegetation index products from the Environment-2 satellite according to claim 2, characterized in that, The lower-order polynomial has an order of 2, and the higher-order polynomial has an order of 3 to 5.

5. The method for producing monthly vegetation index products from the Environment-2 satellite according to claim 1, characterized in that, The preprocessing includes: Sensor system errors were eliminated through radiometric calibration; atmospheric correction was performed using a 6S model to remove atmospheric scattering, absorption, and aerosol interference; and geometric correction was performed through ground control point matching and a quadratic polynomial model.

6. The method for producing monthly vegetation index products from the Environment-2 satellite according to any one of claims 1-5, characterized in that, The method further includes: S7: Execute S4~S6 sequentially according to the natural month to obtain the vegetation index products for each month.

7. The method for producing monthly vegetation index products from the Environment-2 satellite according to claim 6, characterized in that, The method further includes: S8: For vegetation index products for each month, abnormal pixels are removed by combining regional vegetation phenology thresholds, and accuracy is verified based on ground-measured sample points.

8. A device for producing monthly vegetation index products from the Environment-2 satellite, characterized in that, The device includes: The data acquisition module is used to acquire and preprocess remote sensing data from the Environment-2 satellite. The vegetation index calculation module is used to calculate the vegetation index of each period of the Environment-2 satellite remote sensing data to obtain vegetation index images; The maximum value synthesis module is used to synthesize the maximum value of all valid vegetation index images within each ten-day period, based on the natural ten-day period as the basic time unit, to obtain the maximum value image. The 90-month cross-month spatiotemporal reconstruction module is used to acquire the maximum value images of the target month, the month before the target month, and the month after the target month for the target month to be produced, and sort them in chronological order to construct continuous time series data; The sliding filter module is used to seamlessly reconstruct the full-time vegetation index by sliding through the continuous time-series data window by window using the Savitzky-Golay filtering algorithm. The monthly maximum value synthesis module is used to extract the three-ten-day maximum value image corresponding to the target month from the reconstructed continuous time series data, and perform maximum value synthesis to obtain the vegetation index product of the target month.

9. The device for producing monthly vegetation index products from the Environment-2 satellite according to claim 8, characterized in that, The sliding filter module includes: The region division unit is used to calculate the temporal standard deviation of each pixel position based on the continuous time series data, and to divide pixels with a temporal standard deviation greater than a set threshold into high noise regions, and other pixels into low noise regions. A polynomial order setting unit is used to set a low-order polynomial order for the high-noise region and a high-order polynomial order for the low-noise region. The temporal smoothing reconstruction unit is used to construct the least squares fitting equation according to the Savitzky-Golay filtering algorithm, fit the temporal smoothing curve, and use the fitted temporal smoothing curve to interpolate and fill the missing pixels to obtain the temporal reconstruction result of the current round. The least squares fitting equation for the high-noise region uses a low-order polynomial, while the least squares fitting equation for the low-noise region uses a high-order polynomial. The iterative unit is used to calculate the residual between the continuous time series data and the time series reconstruction result of the current round, and uses the 3σ criterion to identify residual noise abnormal observation points and re-mark the abnormal observation points as missing pixels; it returns to the time series smoothing reconstruction unit to repeat the process until the set iteration termination condition is met.

Images