Method for retrieving eucalypt plantation age on monthly scale based on time-series satellite images

Abstract

The present application provides a method for retrieving the age of eucalyptus plantations based on time series satellite images, which belongs to the technical field of forest resource investigation and quantitative remote sensing of forestry. The method can more quickly and accurately obtain the age data of eucalyptus plantations, making up for the shortcomings of traditional methods for obtaining the age of eucalyptus plantations, such as time-consuming, labor-intensive, high cost, and long data acquisition period. The use of long time series remote sensing image vegetation index before and after eucalyptus felling for age retrieval can make up for the shortcomings of the relationship between age and remote sensing spectral reflection to establish an age retrieval model to retrieve the age, and realize the accurate extraction of time series age. By using a spatio-temporal fusion algorithm to spatio-temporally fuse high spatial resolution satellite remote sensing images and low spatial resolution satellite remote sensing images, the missing image data that cannot be used due to cloud and rain in a month is reconstructed, a complete time series image is constructed, and the age of eucalyptus plantations is accurately retrieved to the monthly scale.

Description

Technical Field

[0001] This invention belongs to the field of forest resource survey and quantitative remote sensing technology in forestry, specifically involving a method for inverting the monthly age of eucalyptus plantations based on time-series satellite imagery. Background Technology

[0002] Eucalyptus trees not only grow rapidly and possess excellent wood properties, making them a crucial source of short fiber in pulp and paper production, but they also play a vital role in ecological security, regulating global climate change, and ensuring national timber security. As one of the three major fast-growing tree species, eucalyptus has a short maturation cycle and is typically used primarily for timber production. Therefore, current eucalyptus plantation management mainly focuses on short rotation periods.

[0003] Stand age, as one of the fundamental structural parameters of forests, plays an indispensable role in sustainable forest management and quantitative research on forest resources. In monitoring the dynamic changes of eucalyptus plantation stand volume based on time-series optical imagery, signal saturation easily occurs in optical images under conditions of high biomass or high canopy closure, leading to inaccurate monitoring results. Introducing stand age information as auxiliary data into the eucalyptus plantation stand volume inversion model can improve the accuracy of the dynamic change monitoring results. Stand age information is typically obtained from forest resource inventories, highlighting the urgent need for an advanced method that can quickly, conveniently, and accurately acquire time-series stand age data for eucalyptus plantations.

[0004] With the development of remote sensing technology, some scholars have begun to use satellite remote sensing images to invert forest age. For example, researchers have extracted spectra and vegetation indices from satellite remote sensing images to establish forest age inversion models. Jensen et al. used artificial neural networks to construct a pine forest age inversion model based on TM images; Tang Shaofei et al. used random forests, support vector machines, neural networks, and other algorithms to construct a larch forest age estimation model based on Sentinel-2 images. Although previous research results show that forest age estimation can be achieved based on satellite remote sensing images, the method of establishing a forest age inversion model by using the relationship between remote sensing spectral reflectance and forest age has too many limitations. For example, the spectral reflectance difference between near-mature forests and mature forests of similar age is small, making it difficult to distinguish them accurately. At the same time, the accuracy of forest age inversion is greatly affected by phenological conditions. For example, the accuracy of forest age inversion models at the leaf unfolding stage and leaf fall stage is very different. These factors make it difficult for existing inversion methods based on the correlation between spectral reflectance and forest age to accurately invert the time series forest age of eucalyptus plantations. Therefore, how to accurately extract and promptly update the age information of eucalyptus plantations based on time-series satellite remote sensing images while meeting the needs of short-rotation management data for eucalyptus plantations is a key issue that the management of short-rotation eucalyptus plantation resources aims to address.

[0005] In summary, existing research on eucalyptus plantation age inversion still has the following problems:

[0006] 1. Currently, most eucalyptus plantation age data comes from traditional surveys, which typically involve logging or drilling holes in tree trunks to determine tree age based on tree rings. This method not only severely impacts tree growth but is also time-consuming, labor-intensive, and costly. Furthermore, forest resource inventories are usually conducted every five years, resulting in a low update frequency for eucalyptus plantation age data, making it difficult to meet the needs of short-rotation eucalyptus plantation age data management.

[0007] 2. Optical satellite imagery is greatly affected by clouds and rain, making it difficult to obtain high-quality, continuous time-series image data. This limits the age inversion results based on remote sensing imagery to mostly focusing on the annual scale, making it difficult to further refine the age inversion results to the monthly scale. This hinders the correction of time differences in eucalyptus plantation stock volume at the regional scale, resulting in inaccurate regional stock volume inversion results and posing great difficulties for the precise management of eucalyptus plantation resources.

[0008] 3. The method of establishing a forest age inversion model based on the relationship between remote sensing spectral reflectance and forest age has too many limiting factors. For example, the spectral reflectance difference between near-mature forests and mature forests with similar ages is small, making it difficult to distinguish them accurately. At the same time, the accuracy of forest age inversion is greatly affected by phenological conditions. For example, the accuracy of forest age inversion models at the leaf expansion stage and leaf fall stage is very different. The above factors make it difficult for existing inversion methods based on the correlation between spectral reflectance and forest age to achieve accurate inversion of the time series forest age of eucalyptus plantations. Summary of the Invention

[0009] To address the problems existing in the prior art, this invention provides a method for retrieving the monthly age of eucalyptus plantations based on time-series satellite imagery. The aim is to refine the age retrieval of eucalyptus plantations to a monthly scale, providing more accurate age data to support the management needs of eucalyptus plantations based on short rotation periods.

[0010] The technical solution of this invention is:

[0011] A method for inverting the monthly age of eucalyptus plantations based on time-series satellite imagery, characterized by the following steps:

[0012] Step 1: Determine the target study area for the eucalyptus plantation. Using Google Earth Pro software, through visual interpretation, in historical Google satellite imagery from year Nm to year N: 1) Select multiple pixels containing eucalyptus trees in the study area, i.e., eucalyptus sample points, and record their coordinates; 2) Select multiple pixels corresponding to the felled eucalyptus trees in the study area and record their coordinates and felling update time, defining the felling update time of the eucalyptus trees as the felling update point; where m is a positive integer greater than or equal to the rotation period of eucalyptus trees in the study area;

[0013] Step 2: Select Landsat image data covering the study area that have been orthorectified and georegistered for each year from Nm to N, call these image data on the Google Earth Engine (GEE) platform, and perform image stitching, mask extraction and radiometric normalization preprocessing on these image data to obtain the study area image data for each year from Nm to N.

[0014] Step 3: Extract parameters from the image data of the study area from year Nm to year N. The parameters include spectral reflectance, vegetation index including normalized burn index (NBR), and texture information. Based on the coordinates of eucalyptus sample points, find the corresponding pixels in the image data of the study area from year Nm to year N. Construct a dataset from these pixels and their corresponding parameters. Divide the dataset into a training set and use the random forest algorithm to classify and train eucalyptus trees to construct a spatial distribution extraction model for eucalyptus plantations in the study area for each year. Input the parameters extracted from the image data of the study area for each year into the spatial distribution extraction model for eucalyptus plantations in the study area for each year to obtain the spatial distribution of eucalyptus plantations in the study area from year Nm to year N.

[0015] Step 4: Using the spatial distribution of eucalyptus plantations in the study area from Nm to N years, the corresponding normalized fire index (NBR) spatial distribution obtained from the image data of the study area in each year is masked and extracted to obtain the NBR spatial distribution of eucalyptus plantations in each year, thus obtaining the NBR spatial distribution of eucalyptus plantations in the study area from Nm to N years; the threshold for the sudden drop in NBR is determined based on the change pattern of the NBR value of the corresponding pixel in the NBR spatial distribution of eucalyptus plantations at each logging and regeneration point.

[0016] Step 5: Based on the spatial distribution of NBR in eucalyptus plantations in the Nm to N-year study area and the threshold of NBR drop, determine the actual eucalyptus felling and regeneration points and the corresponding felling and regeneration years in the spatial distribution of eucalyptus plantations in the N-year study area.

[0017] Step 6: Based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the N-year study area, calculate the annual age of eucalyptus plantations in the N-year study area, and then obtain the spatial distribution of annual age of eucalyptus plantations in the N-year study area.

[0018] Step 7: Traverse the Landsat image data for all months of the year corresponding to the nearest felling in the spatial distribution of eucalyptus plantations in the N-year study area for each pixel. After traversal, proceed to Step 11. During the traversal, if the Landsat image data for the current month cannot completely cover the study area and there are missing image data, proceed to Step 8. If the Landsat image data for the current month can completely cover the study area and there are no missing image data, proceed to Step 9.

[0019] Step 8: Introduce MODIS study area image data for the current traversal month and the months before and after it, and obtain the missing image data with the same spatial resolution as the Landsat image data corresponding to the current traversal month through the STARFM spatiotemporal fusion algorithm;

[0020] Step 9: On the GEE platform, perform image stitching, mask extraction, and radiometric normalization preprocessing on the complete image data of the study area for the current traversal month to obtain the preprocessed image data for the current traversal month.

[0021] Step 10: Extract the NBR of the study area image data for the current traversal month directly on the GEE platform. Use the spatial distribution of the eucalyptus plantation in the study area corresponding to the nearest logging and regeneration year N years to the pixel to perform mask extraction on the spatial distribution of the NBR of the study area for the current traversal month, and obtain the spatial distribution of the NBR of the eucalyptus plantation in the study area for the current traversal month. Return to step 7.

[0022] Step 11: Based on the variation of the Normalized Burning Index (NBR) of each pixel in the spatial distribution of eucalyptus plantations in the study area over time within the corresponding month of the nearest logging and regeneration year (N years), the eucalyptus plantation that was just logged in month B of year A is defined as 0 years and 0 months old. Combining this with the spatial distribution of the annual forest age of eucalyptus plantations in the study area over N years, the forest age of the logged eucalyptus trees in the corresponding pixel of the actual logging and regeneration point in the nearest logging and regeneration year A of year N years can be determined as Q years and L months in month C of year N, where Q is the difference between N and A, and L is the difference between C and B.

[0023] Furthermore, according to the method for monthly-scale age inversion of eucalyptus plantations based on time-series satellite imagery, the threshold for the sudden drop in NBR is 0.4.

[0024] Furthermore, according to the method for monthly-scale age inversion of eucalyptus plantations based on time-series satellite imagery, the method for determining the true eucalyptus logging and regeneration points and corresponding logging and regeneration years in step 5 of the study area is as follows: the time points corresponding to the sudden drop in NBR of the pixels in the spatial distribution of eucalyptus plantations in the study area from Nm to N are defined as pseudo logging and regeneration points; then, with the rotation period as the time step, within the time step: if the NBR value of a pixel drops suddenly only once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo logging and regeneration point corresponding to the sudden drop in NBR value is the true logging and regeneration point; if the NBR value of a pixel drops suddenly more than once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo logging and regeneration point corresponding to the last sudden drop in NBR value is taken as the true logging and regeneration point; the current year corresponding to each true logging and regeneration point is defined as the logging and regeneration year of each true logging and regeneration point.

[0025] Furthermore, according to the aforementioned method for retrieving the monthly age of eucalyptus plantations based on time-series satellite imagery, the method for calculating the annual age of eucalyptus plantations in the N-year study area based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the N-year study area is as follows: Eucalyptus trees in each pixel are defined as 0 years old in the logging and regeneration year; the nearest logging and regeneration year to N years is subtracted from N years to obtain the annual age of eucalyptus trees in each pixel in N years; if some pixels in the N-year study area eucalyptus plantations have no logging and regeneration years from Nm to N, then the annual age of eucalyptus trees in those pixels in N years is considered to be greater than m years.

[0026] Compared with the prior art, the technical solution proposed in this invention has the following beneficial effects:

[0027] (1) A more advanced method is provided to obtain eucalyptus plantation age data more quickly and accurately, which makes up for the shortcomings of traditional age acquisition methods, such as being time-consuming, labor-intensive, costly, and having a long data acquisition cycle.

[0028] (2) By using a spatiotemporal fusion algorithm to spatiotemporally fuse high spatial resolution satellite remote sensing images with low spatial resolution satellite remote sensing images, the image data of the missing months that cannot be used due to cloud and rain effects are reconstructed, thereby constructing a complete time series image. In this way, the age inversion results of eucalyptus plantations are further refined to the monthly scale, providing basic data for the correction of time differences in the stock volume of regional eucalyptus plantations.

[0029] (3) Using the changes in vegetation index of long-term remote sensing images before and after eucalyptus felling to invert forest age can make up for the shortcomings of establishing a forest age inversion model based on the relationship between forest age and remote sensing spectral reflectance, and realize the accurate extraction of time series forest age. Attached Figure Description

[0030] To more clearly illustrate the specific methods in the embodiments of the present invention, the relevant drawings involved in the embodiments will be briefly described below. The drawings below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative changes.

[0031] Figure 1 This is a flowchart illustrating the monthly-scale age inversion method for eucalyptus plantations based on time-series satellite imagery in this embodiment.

[0032] Figure 2 This document presents spatial distribution maps of eucalyptus plantations in the study area from 2006 to 2020 for each year of this implementation method. Map (a) shows the spatial distribution of eucalyptus plantations in the study area in 2006; (b) shows the spatial distribution in 2007; (c) shows the spatial distribution in 2008; (d) shows the spatial distribution in 2009; (e) shows the spatial distribution in 2010; (f) shows the spatial distribution in 2011; and (g) shows the spatial distribution in 2012. Spatial distribution maps: (h) Spatial distribution map of eucalyptus plantations in the study area in 2013; (i) Spatial distribution map of eucalyptus plantations in the study area in 2014; (j) Spatial distribution map of eucalyptus plantations in the study area in 2015; (k) Spatial distribution map of eucalyptus plantations in the study area in 2016; (l) Spatial distribution map of eucalyptus plantations in the study area in 2017; (m) Spatial distribution map of eucalyptus plantations in the study area in 2018; (n) Spatial distribution map of eucalyptus plantations in the study area in 2019; (o) Spatial distribution map of eucalyptus plantations in the study area in 2020.

[0033] Figure 3 (a) is a schematic diagram of the pixels represented by black squares corresponding to a certain logging and renewal point in this embodiment; (b) is a graph showing the change of the normalized combustion index (NBR) of the logging and renewal point corresponding to the pixels represented by black squares in (a) in the logging and renewal year and the years before and after.

[0034] Figure 4 This is a spatial distribution map of the age of eucalyptus plantations in the study area in 2020, based on this implementation method.

[0035] Figure 5 This is a spatial distribution map of the monthly standing age of eucalyptus plantations in the study area in 2020, based on this implementation method. Detailed Implementation

[0036] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings. Preferred embodiments of this application are shown in the drawings. However, this application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of this application.

[0037] This embodiment uses the inversion of the monthly age of eucalyptus plantations in Guangxi Zhuang Autonomous Region in 2020 based on Landsat series image data from 2006 to 2020 as an example to describe in detail the method for inverting the monthly age of eucalyptus plantations based on time-series satellite imagery of the present invention. Figure 1 This is a flowchart illustrating the monthly-scale age inversion method for eucalyptus plantations based on time-series satellite imagery, as described in this embodiment. Figure 1 As shown, the method includes the following steps:

[0038] Step 1: Determine the target study area for the eucalyptus plantation. Using Google Earth Pro software, through visual interpretation, in historical Google satellite imagery from year Nm to year N: 1) Select multiple pixels containing eucalyptus trees in the study area, i.e., eucalyptus sample points, and record their coordinates; 2) Select multiple pixels corresponding to the felled eucalyptus trees in the study area and record their coordinates and felling update time, defining the felling update time of the eucalyptus trees as the felling update point; where m is a positive integer greater than or equal to the rotation period of eucalyptus trees in the study area;

[0039] The target study area for eucalyptus plantations identified in this implementation method is the Guangxi Zhuang Autonomous Region, where the rotation period for eucalyptus trees is typically 3-5 years. This implementation method utilizes Google Earth Pro software to visually interpret historical Google satellite imagery from 2006-2020 (the year span must be long enough than the eucalyptus rotation period to accurately determine the stand age) and select 1000 eucalyptus sample points distributed throughout the study area, recording their coordinates. Then, using visual interpretation, 600 eucalyptus felling and replanting points are randomly selected from the historical Google satellite imagery from 2006-2020, and their corresponding pixel coordinates and felling and replanting times are recorded.

[0040] In this step, historical Google satellite imagery from 2006 to 2020 was opened in Google Earth Pro software. The location of the Guangxi Zhuang Autonomous Region was then identified within the imagery. Since eucalyptus trees are mostly plantations with clear spacing between trees, the texture of these plantations is evident in the historical Google satellite imagery. Using visual interpretation, 1000 eucalyptus sample points, evenly distributed throughout the Guangxi Zhuang Autonomous Region, were selected from each year's historical Google satellite imagery, totaling 15,000 sample points. The coordinates of each sample point were recorded. Visual interpretation was then used to analyze the historical Google satellite imagery from 2006 to 2020. Sixty hundred pixels corresponding to felled eucalyptus trees distributed within the Guangxi Zhuang Autonomous Region were selected from historical satellite imagery. First, areas with concentrated eucalyptus distribution were identified in Google historical satellite imagery from 2006 to 2020. Using visual interpretation, the pixels corresponding to the felled eucalyptus trees were located. A portion of these pixels was located each year from 2006 to 2020, totaling 600 pixels, and the coordinates of each pixel were recorded. Each pixel corresponds to a eucalyptus felling and regeneration point. By comparing the historical satellite imagery of the selected 600 pixels with those of the preceding and following years, the felling and regeneration time of the eucalyptus trees within each pixel was determined and recorded. The felling and regeneration time of the eucalyptus trees within each pixel is defined as the eucalyptus felling and regeneration point corresponding to that pixel.

[0041] Step 2: Select Landsat image data covering the study area that have been orthorectified and georegistered for each year from Nm to N, call these image data on the Google Earth Engine (GEE) platform, and perform image stitching, mask extraction, and radiometric normalization preprocessing on these image data to obtain the study area image data for each year from Nm to N.

[0042] This implementation selects the latest Landsat imagery data covering the study area that has been orthorectified and georegistered for each year from 2006 to 2020. This imagery data is then accessed on the Google Earth Engine (GEE) platform, and preprocessed, including image stitching, mask extraction, and radiometric normalization, to obtain preprocessed imagery data of the study area for each year from 2006 to 2020.

[0043] In the specific implementation of this method, considering the launch times of the Landsat series satellites, Landsat-5 image data from 2006-2011, Landsat-7 image data from 2012, and Landsat-8 image data from 2013-2020 were selected. This image data was directly accessed on the GEE platform. The image data for a specific year's study area was obtained by stitching together multiple Landsat images from different months of that year. For example, using Landsat-5 image data from different months of 2006 that completely cover the study area, the vector boundary map of the study area (which can be downloaded online) was stitched together. Then, mask extraction was performed using the vector boundary map to extract the portion within the study area from the stitched image, discarding the portion outside the study area to obtain the 2006 study area image data. This process was repeated to obtain study area image data for each year from 2007 to 2020. Because there are certain differences in surface reflection among the images acquired by the TM sensor carried by Landsat-5, the ETM+ sensor carried by Landsat-7, and the OLI sensor carried by Landsat-8, radiometric normalization processing is required. The radiometric normalization algorithm built into GEE is used for this processing. The image data of the study area from 2006 to 2020 can be directly input into the radiometric normalization algorithm to obtain the radiometrically normalized image data of the study area for each year from 2006 to 2020.

[0044] Step 3: Extract parameters from the image data of the study area for each year. The parameters include spectral reflectance, vegetation index including normalized burn index (NBR), and texture information. Based on the coordinates of eucalyptus sample points, find the corresponding pixels in the image data of the study area from year Nm to year N. Construct a dataset from these pixels and their corresponding parameters. Divide the dataset into a training set and use the random forest algorithm to classify and train eucalyptus trees to construct a spatial distribution extraction model for eucalyptus plantations in the study area for each year. Input the parameters extracted from the image data of the study area for each year into the spatial distribution extraction model for eucalyptus plantations in the study area for each year to obtain the spatial distribution of eucalyptus plantations in the study area from year Nm to year N.

[0045] This implementation method extracts parameters from preprocessed image data of the study area from 2006 to 2020, including spectral reflectance, vegetation index, and texture information. Using the eucalyptus sample data collected in step 1, a dataset is constructed using ArcGIS software. A random forest algorithm is employed for classification, and a spatial distribution extraction model for eucalyptus plantations is built. Based on the preprocessed image data of the study area from 2006 to 2020, the spatial distribution of eucalyptus plantations in the study area for each year from 2006 to 2020 is obtained.

[0046] During this step, the spectral reflectance, vegetation index, and texture information of the preprocessed image data of the study area from 2006 to 2020 can be directly extracted on the GEE platform. Among them, spectral reflectance is the ratio of the light flux reflected by different ground objects in different bands by the satellite-borne sensor to the light flux incident on the object; vegetation indices include: Enhanced Vegetation Index (EVI), Normalized Difference Vegetation Index (NDVI), Normalized Difference Vegetation Index (NDVIA), Terrestrial Chlorophyll Index (MTCI), Novel Inverted Red Edge Chlorophyll Index (IRECI), Vegetation Attenuation Index (PSRI), Transformed Chlorophyll Absorbed Reflectance Index (TCARI), Normalized Difference Water Index (NDWI), Modified Chlorophyll Absorbed Reflectance Index (MCARI), Renormalized Vegetation Index (RDVI), Triangular Vegetation Index (TVI), Soil-Regulated Vegetation Index (SAVI), Water Stress Index (MSI), Surface Water Index (LSWI), Normalized Burning Index (NBR), Enhanced Normalized Red Edge Vegetation Index (mNdvired_edge), Improved Red Edge Ratio Vegetation Index (MSRred_edge), and Chlorophyll Red Edge Index (CIred_edge); texture information includes: mean difference, variance, homogeneity, contrast, dissimilarity, information entropy, second moment, and correlation.

[0047] The following example illustrates the process of obtaining the spatial distribution of eucalyptus plantations in the study area for each year: The preprocessed 2006 image data of the study area is opened in ArcGIS. The coordinates of the 1000 eucalyptus sample points obtained in step 1 are input into ArcGIS, allowing the corresponding points to be found in the preprocessed 2006 image data. Since the preprocessed image data of the study area from 2006 to 2020 are all raster images, the 1000 eucalyptus sample points are mapped to the 2006 image data. Using ArcGIS's raster image multi-value extraction to point function, the spectral reflectance, vegetation index, and texture information of the corresponding pixels extracted from the 2006 image data are extracted into the 1000 eucalyptus sample points. Therefore, the attribute tables of each of the 1000 eucalyptus sample points contain their corresponding spectral reflectance, vegetation index, and texture information. Using the classification results of 1000 eucalyptus sample points as eucalyptus, and the spectral reflectance, vegetation index, and texture information corresponding to the attribute tables of each of the 1000 eucalyptus sample points as features, a dataset for extracting the spatial distribution of eucalyptus plantations in the study area in 2006 was constructed. 75% of the data in the dataset was used as the training set, and the remaining 25% as the validation set. The random forest algorithm was used for eucalyptus classification training, thus constructing the spatial distribution extraction model for eucalyptus plantations in the study area in 2006. Then, the 2006 image data of the study area was input into the constructed spatial distribution extraction model for eucalyptus plantations in the study area in 2006. The random forest algorithm, based on the extracted features (spectral reflectance, vegetation index, and texture information) of the 2006 image data of the study area, can obtain the spatial distribution of eucalyptus plantations in the study area in 2006. The spatial distribution of eucalyptus plantations in the study area for each year from 2006 to 2020 is presented as raster images, with eucalyptus trees forming the spatial distribution of eucalyptus plantations in the form of pixels.

[0048] By performing the above operations on the image data of each year in the study area from 2007 to 2020, the spatial distribution model of eucalyptus plantations in the study area from 2007 to 2020 can be obtained, thus yielding the spatial distribution of eucalyptus plantations in the study area from 2006 to 2020. Figure 2The image shows the spatial distribution maps of eucalyptus plantations in the study area from 2006 to 2020. (a) shows the spatial distribution maps of eucalyptus plantations in the study area in 2006, (b) in 2007, (c) in 2008, (d) in 2009, (e) in 2010, (f) in 2011, and (g) in 2012. Distribution maps: (h) shows the spatial distribution of eucalyptus plantations in the study area in 2013; (i) shows the spatial distribution of eucalyptus plantations in the study area in 2014; (j) shows the spatial distribution of eucalyptus plantations in the study area in 2015; (k) shows the spatial distribution of eucalyptus plantations in the study area in 2016; (l) shows the spatial distribution of eucalyptus plantations in the study area in 2017; (m) shows the spatial distribution of eucalyptus plantations in the study area in 2018; (n) shows the spatial distribution of eucalyptus plantations in the study area in 2019; and (o) shows the spatial distribution of eucalyptus plantations in the study area in 2020.

[0049] Step 4: Using the spatial distribution of eucalyptus plantations in the study area from Nm to N years, the corresponding normalized fire index (NBR) spatial distribution obtained from the image data of the study area in each year is masked and extracted to obtain the NBR spatial distribution of eucalyptus plantations in each year, thus obtaining the NBR spatial distribution of eucalyptus plantations in the study area from Nm to N years; the threshold for the sudden drop in NBR is determined based on the change pattern of the NBR value of the corresponding pixel in the NBR spatial distribution of eucalyptus plantations at each logging and regeneration point.

[0050] In this step, the spatial distribution of eucalyptus plantations in the study area from 2006 to 2020, obtained in step 3, and the Normalized Burning Index (NBR) extracted from the image data of the study area from 2006 to 2020, are opened in ArcGIS software. The extracted NBR for each year of 2006 to 2020 exists in the form of spatial distribution, that is, the spatial distribution of NBR for each year of 2006 to 2020 can be obtained from step 3. Using the mask extraction function in ArcGIS software, the spatial distribution of eucalyptus plantations in the study area from 2006 to 2020 is used to mask the spatial distribution of NBR extracted from the image data of the study area from 2006 to 2020, respectively, to obtain the spatial distribution of NBR for each year of eucalyptus plantations, thus obtaining the spatial distribution of NBR for eucalyptus plantations in the study area from 2006 to 2020. In ArcGIS, the 600 logging update points obtained in step 1 were expanded. Each logging update point corresponds to the pixel value of the corresponding location in the spatial distribution of the Normalized Burning Index (NBR) of eucalyptus plantations in the study area from 2006 to 2020. The changes in the NBR values ​​of the corresponding pixels in the spatial distribution of the NBR of eucalyptus plantations in the study area from 2006 to 2020 were observed. It was found that the NBR values ​​of the pixels corresponding to each eucalyptus logging update point showed a change of "stable-sharp drop-rise-stable". Compared with other years within the eucalyptus rotation period, the NBR value of the current year corresponding to the logging update point dropped rapidly. Figure 3 The changes in the NBR values ​​of the pixels shown in Figure (a) are illustrated in Figure (b). The NBR values ​​corresponding to the logging and regeneration points in 2016 dropped rapidly. In this embodiment, the drop in the normalized rate of combustion (NBR) of most pixels corresponding to logging and regeneration points is around 0.45. Taking the minimum drop value of 0.4, this embodiment sets the threshold for the drop in NBR to 0.4.

[0051] Step 5: Based on the spatial distribution of NBR in eucalyptus plantations in the Nm to N-year study area and the threshold of NBR drop, determine the actual eucalyptus felling and regeneration points and the corresponding felling and regeneration years in the spatial distribution of eucalyptus plantations in the N-year study area.

[0052] The method is as follows: The time point corresponding to the sudden drop in NBR of the pixels in the spatial distribution of eucalyptus plantations in the study area from Nm to N years is defined as the pseudo-logging and renewal point of eucalyptus; then, with the rotation period as the time step, within the time step: if the NBR value of a pixel drops suddenly only once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo-logging and renewal point corresponding to that NBR value drop is the real logging and renewal point; if the NBR value of a pixel drops suddenly more than once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo-logging and renewal point corresponding to the last NBR value drop is taken as the real logging and renewal point; the current year corresponding to each real logging and renewal point is defined as the logging and renewal year of each real logging and renewal point.

[0053] Based on the spatial distribution of the Normalized Burning Index (NBR) of eucalyptus plantations in the study area from 2006 to 2020, the time points corresponding to the sudden drop in NBR of the pixels in the spatial distribution of eucalyptus plantations in the study area in each year from 2006 to 2020 were defined as eucalyptus pseudo-felling and regeneration points. Based on the preset time step and the threshold of the sudden drop in NBR, pseudo-felling was removed, and the real eucalyptus felling and regeneration points and the corresponding felling and regeneration years in the spatial distribution of eucalyptus plantations in the study area in 2020 were determined.

[0054] In this step, the time points corresponding to the sudden drop in NBR of pixels in the spatial distribution of eucalyptus plantations in the study area from 2006 to 2020 are defined as pseudo-eucalyptus logging and renewal points, and the current year corresponding to each pseudo-eucalyptus logging and renewal point is defined as the logging and renewal year of the pseudo-eucalyptus logging and renewal point. In reality, when no logging event occurs in the area where a pixel is located, the Normalized Burning Index (NBR) will also decrease or increase. For example, due to the different growth conditions of eucalyptus plantations each year, the NBR will increase when there is sufficient rainfall and sunlight, and decrease when there is drought or pests and diseases. This invention uniformly defines the time points corresponding to the decrease in NBR of pixels as pseudo-logging and renewal points, and then performs a pseudo-logging removal process to eliminate erroneous logging and renewal points caused by factors other than logging, obtaining the true logging and renewal points. Considering the specific circumstances of eucalyptus plantation logging, since eucalyptus is generally logged once every 5 years, this implementation method presets a time step of 5 years. If the Normalized Burning Index (NBR) of a pixel drops sharply only once within a total of 5 years, and the magnitude of the drop is greater than or equal to the threshold for the magnitude of the drop in NBR, then the pseudo-logging update point corresponding to that pixel is the true logging update point, and the current year corresponding to the drop is the logging update year of that true logging update point. If the NBR of a pixel drops sharply more than once within a total of 5 years, and the magnitude of each drop is greater than or equal to the threshold for the magnitude of the drop in NBR, then the pseudo-logging update point corresponding to the last NBR drop is taken as the true logging update point, and the current year corresponding to that drop is the logging update year of the true logging update point. Other pseudo-logging update points are considered erroneous logging update points.

[0055] In summary, based on the spatial distribution of eucalyptus plantations in the study area in 2020, the true logging and regeneration points can be obtained by subtracting the erroneous logging and regeneration points from the false logging and regeneration points. Each true logging and regeneration point corresponds to a logging and regeneration year. The logging and regeneration years corresponding to each pixel in the spatial distribution of eucalyptus plantations in the study area in 2020 were statistically analyzed.

[0056] Step 6: Based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the N-year study area, calculate the annual age of eucalyptus plantations in the N-year study area, and then obtain the spatial distribution of annual age of eucalyptus plantations in the N-year study area.

[0057] The method for calculating the age of eucalyptus plantations in the N-year study area is as follows: Define the age of eucalyptus trees in each pixel as 0 years old in the year of felling and regeneration. Subtract the nearest year of felling and regeneration to N years from N years to obtain the age of eucalyptus trees in each pixel in N years. If some pixels in the eucalyptus plantations in the N-year study area have no years of felling and regeneration between Nm and N years, then the age of eucalyptus trees in those pixels in N years is considered to be greater than m years.

[0058] In this embodiment, based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the study area in 2020, obtained in step 5, the annual age of eucalyptus plantations in the study area in 2020 is calculated, thereby obtaining the spatial distribution of the annual age of eucalyptus plantations in the study area in 2020. During the implementation of this step, the annual age of eucalyptus plantations in the study area in 2020 is calculated by programming in the GEE platform. The programming approach is as follows: based on the pixels in the spatial distribution of eucalyptus plantations in the study area in 2020 (the spatial distribution of eucalyptus plantations in the study area in 2020 is a raster image, therefore eucalyptus plantations exist in the form of pixels), the logging and regeneration year of the actual logging and regeneration point corresponding to each pixel is queried, and the nearest logging and regeneration year of the eucalyptus trees in each pixel to N years ago is determined. Since the eucalyptus trees within the corresponding pixels of the actual logging-renewal points were logged in the corresponding logging-renewal year, the age of the eucalyptus trees in each pixel is defined as 0 years old in the logging-renewal year. Therefore, after determining the nearest logging-renewal year N years from each pixel, the age of the eucalyptus trees in each pixel is obtained by subtracting the nearest logging-renewal year N years from 2020. For example, if the nearest logging-renewal year from 2020 for a certain pixel in the spatial distribution of eucalyptus plantations in the study area is 2015, then the eucalyptus trees in that pixel were 0 years old in 2015 and 5 years old in 2020. If some pixels in the eucalyptus plantations in the study area had no logging-renewal years from 2006 to 2020, then they are considered to be older than 14 years old in 2020. In summary, the annual stand age of eucalyptus trees corresponding to all pixels in the spatial distribution of eucalyptus plantations in the study area in 2020 can be obtained, thus yielding the spatial distribution of annual stand age of eucalyptus plantations in the study area in 2020. The spatial distribution of annual stand age of eucalyptus plantations in the study area for each year from 2007 to 2019 can also be obtained using the above method. However, since this implementation method did not use image data prior to 2006, the annual stand age of eucalyptus plantations in the study area in 2006 cannot be retrieved. Figure 4 The figure shows the spatial distribution of the age of eucalyptus plantations in the study area in 2020.

[0059] Step 7: Traverse the Landsat image data for all months of the year corresponding to the nearest felling in the spatial distribution of eucalyptus plantations in the N-year study area for each pixel. After traversal, proceed to Step 11. During the traversal, if the Landsat image data for the current month cannot completely cover the study area and there are missing image data, proceed to Step 8. If the Landsat image data for the current month can completely cover the study area and there are no missing image data, proceed to Step 9.

[0060] In this embodiment, based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the study area in 2020, as statistically determined in step 5, the nearest logging and regeneration year to 2020 corresponding to each pixel in the spatial distribution of eucalyptus plantations in 2020 can be determined. The Landsat-8 image data for all months of the nearest logging and regeneration year to 2020 corresponding to each pixel in the study area in 2020 are traversed. If the Landsat-8 image data for the currently traversed month can completely cover the study area, then step 9 is executed; if the Landsat-8 image data for the currently traversed month cannot completely cover the study area, then step 8 is executed.

[0061] Step 8: Introduce MODIS study area image data of the currently traversed month and the months before and after it, and obtain the missing image data with the same spatial resolution as the Landsat image data corresponding to the currently traversed month through the STARFM spatiotemporal fusion algorithm.

[0062] In this embodiment, Landsat-8 imagery data has high spatial resolution but low temporal resolution, with a revisit period of 16 days. Due to weather conditions, satellite imagery data may sometimes be affected by clouds, rendering it unusable. Therefore, only two Landsat-8 satellite images are available for the same area in the same month, and cloud-affected images may not be usable. MODIS imagery data has high temporal resolution and a replay period of 1 day, with one satellite image available for the same area in the same month each day. However, MODIS imagery has low spatial resolution, resulting in relatively poor forest age inversion. Therefore, spatiotemporal fusion of MODIS and Landsat-8 imagery data can compensate for the missing Landsat-8 imagery data in the current month of revisiting. For example, in the spatial distribution of eucalyptus plantations in the study area in 2020, the nearest logging and regeneration year to a certain pixel is 2020. However, the Landsat-8 imagery from April 2020 cannot fully cover the study area. By inputting the Landsat-8 imagery from March 31, 2020, the MODIS imagery from March 31, 2020, the Landsat-8 imagery from May 2, 2020, the MODIS imagery from May 2, 2020, and the MODIS imagery from April 15, 2020 into the STARFM spatiotemporal fusion algorithm, imagery data from April 2020 with the same spatial resolution as Landsat-8 that can fully cover the study area can be obtained. Using the method described above, complete image data of the study area in 2020 that can fully cover the study area in each month of 2020 were obtained. This yielded complete image data of each pixel in the spatial distribution of eucalyptus plantations in the study area that can fully cover the study area in each month of the year closest to the logging and regeneration in 2020.

[0063] Step 9: On the GEE platform, perform image stitching, mask extraction, and radiometric normalization preprocessing on the complete image data of the study area for the current traversal month to obtain the preprocessed image data for the current traversal month.

[0064] In this embodiment, the complete image data of the study area for the current traversal month within the current logging and regeneration year for corresponding pixels is preprocessed using the GEE platform through image stitching, mask extraction, and radiometric normalization. This yields preprocessed pre-image data for each pixel in the spatial distribution of eucalyptus plantations in the study area for the current traversal month of the most recent logging and regeneration year in 2020. During this step, the complete image data of the study area for the current traversal month within the most recent logging and regeneration year for corresponding pixels is opened on the GEE platform and image stitched. Mask extraction is performed using the study area vector boundary map to remove the portion outside the vector boundary. Then, the radiometric normalization algorithm built into GEE is used for processing to obtain preprocessed image data for each pixel in the spatial distribution of eucalyptus plantations in the study area for the current month of the most recent logging and regeneration year in 2020.

[0065] Step 10: Extract the NBR of the study area image data for the current traversal month directly on the GEE platform. Use the spatial distribution of the eucalyptus plantation in the study area corresponding to the nearest logging and regeneration year N years to the pixel to perform mask extraction on the spatial distribution of the NBR of the study area for the current traversal month, and obtain the spatial distribution of the NBR of the eucalyptus plantation in the study area for the current traversal month. Return to step 7.

[0066] In this embodiment, the Normalized Burn Index (NBR) of the study area image data corresponding to the most recent logging and regeneration year in 2020 for each pixel in the spatial distribution of eucalyptus plantations in the study area is directly extracted on the GEE platform. The NBR of the study area image data corresponding to the most recent logging and regeneration year in 2020 for each pixel in the spatial distribution of eucalyptus plantations in the study area exists in the form of spatial distribution, that is, the spatial distribution of the NBR of the study area NBR corresponding to the most recent logging and regeneration year in 2020 for each pixel in the spatial distribution of eucalyptus plantations in the study area. Using the spatial distribution of eucalyptus plantations in the study area in 2020, corresponding to the nearest logging and regeneration year for each pixel, the Normalized Burning Index (NBR) spatial distribution of the corresponding pixel for the current month of the logging and regeneration year is masked and extracted. This yields the NBR spatial distribution of eucalyptus plantations in the study area in 2020 for each pixel in the spatial distribution of eucalyptus plantations in the study area for the current month of the nearest logging and regeneration year. Return to step 7 until the NBR spatial distribution of eucalyptus plantations in the study area for all months of the nearest logging and regeneration year in 2020 is obtained. For example, if a pixel in 2020 corresponds to a eucalyptus logging and regeneration year of 2015, and Landsat imagery data for one or more months in 2015 does not fully cover the study area, then through steps 7 to 10, the spatial distribution of the Normalized Burning Index (NBR) of the eucalyptus plantations in the study area for each month in 2015 can be obtained. Similarly, the spatial distribution of the NBR of the eucalyptus plantations in the study area for each month corresponding to the nearest logging and regeneration year to the actual logging and regeneration point corresponding to each pixel in the spatial distribution of the eucalyptus plantations in the study area in 2020 can be obtained.

[0067] Step 11: Based on the variation of the Normalized Burning Index (NBR) of each pixel in the spatial distribution of eucalyptus plantations in the study area over time within the corresponding month of the nearest logging and regeneration year (N years), the eucalyptus plantation that was just logged in month B of year A is defined as 0 years and 0 months old. Combining this with the spatial distribution of the annual forest age of eucalyptus plantations in the study area over N years, the forest age of the logged eucalyptus trees in the corresponding pixel of the actual logging and regeneration point in the nearest logging and regeneration year A of year N years can be determined as Q years and L months in month C of year N, where Q is the difference between N and A, and L is the difference between C and B.

[0068] In this embodiment, by observing the change of the Normalized Burning Index (NBR) of each pixel in the spatial distribution of eucalyptus plantations in the study area in each month of the corresponding year closest to the logging and regeneration year of 2020, the NBR value of eucalyptus plantations is the lowest when they are first logged. As the eucalyptus trees grow, their NBR gradually increases and then tends to level off. Therefore, the newly logged eucalyptus plantations are taken as 0 years and 0 months. For example, in the spatial distribution of eucalyptus plantations in the study area in 2020, the nearest logging and regeneration year to 2020 for a certain pixel is 2015. Based on the spatial distribution of the Normalized Burning Index (NBR) of eucalyptus plantations in the study area corresponding to the nearest logging and regeneration year to 2020 for each pixel in the spatial distribution of eucalyptus plantations in the study area, the spatial distribution of the NBR of eucalyptus plantations in the study area for each month in 2015 can be obtained. By observing the change pattern of the Normalized Burning Index (NBR) value of this pixel within the 12 months of 2015, it is found that the NBR value of this pixel is lowest in May of 2015. Therefore, the eucalyptus plantations within this pixel... In May 2015, the age was 0 years and 0 months. Based on the month of logging, the monthly age of eucalyptus plantations was calculated cumulatively. Thus, the eucalyptus plantation in this pixel was 0 years and 7 months old in December 2015. Combining this with the spatial distribution of the annual age of eucalyptus plantations in the study area in 2020, it was calculated that the eucalyptus plantation in this pixel was 5 years and 7 months old in December 2020. By analogy, the annual age of eucalyptus plantations in all pixels of the spatial distribution of eucalyptus plantations in the study area in 2020 can be refined to the monthly age, thus obtaining the spatial distribution of the monthly age of eucalyptus plantations in the study area in 2020. This allows the annual age of eucalyptus plantations in 2020 to be accurately retrieved to the monthly scale. The spatial distribution of monthly tree age of eucalyptus plantations in the study area from 2007 to 2019 can also be obtained using the above method. However, since this implementation method did not use image data prior to 2006, the annual tree age of eucalyptus plantations in the study area in 2006 cannot be retrieved, and therefore the monthly tree age of eucalyptus plantations in the study area in 2006 cannot be retrieved. Figure 5 The figure shows the spatial distribution of monthly tree age of eucalyptus plantations in the study area in 2020. In the legend, x / x represents x years and x months, for example, 1 / 01 represents 1 year and 1 month.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; therefore, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the claims of the present invention.

Claims

1. A method for monthly-scale age inversion of eucalyptus plantations based on time-series satellite imagery, characterized in that, The method includes the following steps: Step 1: Determine the target study area for the eucalyptus plantation. Using Google Earth Pro software, through visual interpretation, in historical Google satellite imagery from year Nm to year N: 1) Select multiple pixels containing eucalyptus trees in the study area, i.e., eucalyptus sample points, and record their coordinates; 2) Select multiple pixels corresponding to the felled eucalyptus trees in the study area and record their coordinates and felling renewal time, defining the felling renewal time of the eucalyptus trees as the felling renewal point; where m is a positive integer greater than or equal to the rotation period of eucalyptus trees in the study area; Step 2: Select Landsat image data covering the study area that have been orthorectified and georegistered for each year from Nm to N, call these image data on the Google Earth Engine (GEE) platform, and perform image stitching, mask extraction and radiometric normalization preprocessing on these image data to obtain the study area image data for each year from Nm to N. Step 3: Extract parameters from the image data of the study area from year Nm to year N. The parameters include spectral reflectance, vegetation index including normalized burn index (NBR), and texture information. Based on the coordinates of eucalyptus sample points, find the corresponding pixels in the image data of the study area from year Nm to year N. Construct a dataset from these pixels and their corresponding parameters. Divide the dataset into a training set and use the random forest algorithm to classify and train eucalyptus trees to construct a spatial distribution extraction model for eucalyptus plantations in the study area for each year. Input the parameters extracted from the image data of the study area for each year into the spatial distribution extraction model for eucalyptus plantations in the study area for each year to obtain the spatial distribution of eucalyptus plantations in the study area from year Nm to year N. Step 4: Using the spatial distribution of eucalyptus plantations in the study area from Nm to N years, the corresponding normalized fire index (NBR) spatial distribution obtained from the image data of the study area in each year is masked and extracted to obtain the NBR spatial distribution of eucalyptus plantations in each year, thus obtaining the NBR spatial distribution of eucalyptus plantations in the study area from Nm to N years; the threshold for the sudden drop in NBR is determined based on the change pattern of the NBR value of the corresponding pixel in the NBR spatial distribution of eucalyptus plantations at each logging and regeneration point. Step 5: Based on the spatial distribution of NBR in eucalyptus plantations in the Nm to N-year study area and the threshold of NBR drop, determine the actual eucalyptus felling and regeneration points and the corresponding felling and regeneration years in the spatial distribution of eucalyptus plantations in the N-year study area. Step 6: Based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the N-year study area, calculate the annual age of eucalyptus plantations in the N-year study area, and thus obtain the spatial distribution of annual age of eucalyptus plantations in the N-year study area. Step 7: Traverse the Landsat image data for all months of the year corresponding to the nearest felling in the spatial distribution of eucalyptus plantations in the N-year study area for each pixel. After traversal, proceed to Step 11. During the traversal, if the Landsat image data for the current month cannot completely cover the study area and there are missing image data, proceed to Step 8. If the Landsat image data for the current month can completely cover the study area and there are no missing image data, proceed to Step 9. Step 8: Introduce MODIS study area image data for the current traversal month and the months before and after it, and obtain the missing image data with the same spatial resolution as the Landsat image data corresponding to the current traversal month through the STARFM spatiotemporal fusion algorithm; Step 9: On the GEE platform, perform image stitching, mask extraction, and radiometric normalization preprocessing on the complete image data of the study area for the current traversal month to obtain the preprocessed image data for the current traversal month. Step 10: Extract the NBR of the study area image data for the current traversal month directly on the GEE platform. Use the spatial distribution of the eucalyptus plantation in the study area corresponding to the nearest logging and regeneration year N years to the pixel to perform mask extraction on the spatial distribution of the NBR of the study area for the current traversal month, and obtain the spatial distribution of the NBR of the eucalyptus plantation in the study area for the current traversal month. Return to step 7. Step 11: Based on the variation of the Normalized Burning Index (NBR) of each pixel in the spatial distribution of eucalyptus plantations in the study area over time in the corresponding months of the nearest logging and regeneration year (N years away), the eucalyptus plantation that was just logged in month B of year A is defined as 0 years and 0 months old. Combining this with the spatial distribution of the annual forest age of eucalyptus plantations in the study area over N years, the forest age of the logged eucalyptus trees in the corresponding pixel of the actual logging and regeneration point in the nearest logging and regeneration year A of year N years can be determined as Q years and L months in month C of year N, where Q is the difference between N and A, and L is the difference between C and B. The method for determining the actual eucalyptus logging and regeneration points and corresponding logging and regeneration years in the spatial distribution of eucalyptus plantations in the N-year study area in step 5 is as follows: the time point corresponding to the sudden drop in NBR of the pixels in the spatial distribution of eucalyptus plantations in the study area from Nm to N years is defined as the pseudo logging and regeneration point; then, with the rotation period as the time step, within the time step: if the NBR value of a pixel drops suddenly only once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo logging and regeneration point corresponding to the sudden drop in NBR value is the actual logging and regeneration point; if the NBR value of a pixel drops suddenly more than once, and the magnitude of the drop is greater than or equal to the NBR drop magnitude threshold, then the pseudo logging and regeneration point corresponding to the last sudden drop in NBR value is taken as the actual logging and regeneration point; the current year corresponding to each actual logging and regeneration point is defined as the logging and regeneration year of each actual logging and regeneration point.

2. The method for monthly-scale age inversion of eucalyptus plantations based on time-series satellite imagery according to claim 1, characterized in that, The threshold for the sudden drop in NBR is 0.

4.

3. The method for monthly-scale age inversion of eucalyptus plantations based on time-series satellite imagery according to claim 1, characterized in that, Based on the logging and regeneration year corresponding to each pixel in the spatial distribution of eucalyptus plantations in the N-year study area, the method for calculating the annual age of eucalyptus plantations in the N-year study area is as follows: Define the age of eucalyptus trees in each pixel as 0 years old in the logging and regeneration year; subtract the nearest logging and regeneration year to N years from N years to obtain the annual age of eucalyptus trees in each pixel in N years; if some pixels in the eucalyptus plantations in the N-year study area have no logging and regeneration years from Nm to N years, then the annual age of eucalyptus trees in these pixels in N years is considered to be greater than m years old.

Images