Rubber plantation distribution and establishment time recognition method based on full-time sequence images

Abstract

The application discloses a rubber plantation distribution and establishment time recognition algorithm based on full-time sequence images, and comprises the following steps: step 1, data acquisition and pretreatment; step 2, time sequence analysis and time correction to determine the planting year based on LandTrendr; step 3, extraction of key features in the economic life cycle; step 4, random forest classification and identification of rubber; and step 5, precision verification and evaluation. The application realizes accurate identification of the spatial distribution and establishment year of the rubber plantation by analyzing the key features in the entire economic life cycle of the rubber plantation. The method has strong adaptability and generalizability, and has significant identification advantages under complex landscape conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote sensing image processing, and in particular to a rubber plantation distribution and establishment time recognition method based on full-time sequence images. BACKGROUND

[0002] It is an urgent need to accurately monitor the spatio-temporal dynamics of rubber plantations using remote sensing to promote the sustainable development of the rubber industry. With the development of space-based observation technology, remote sensing methods play an increasingly important role in the monitoring of rubber plantations. Although remote sensing monitoring of rubber plantations has experienced rapid development, the current mainstream algorithms are based on rubber forests in the northern tropics, and the application scenarios are generally concentrated at the provincial and municipal scales. In the application process of large-scale rubber plantation distribution and change detection, there are significant inconsistencies. These problems limit the comprehensive and objective understanding of the dynamics of rubber plantations in the current core planting areas and the impact of rubber plantations on the regional ecological environment. Specifically, the existing technology mainly has the following problems:

[0003] 1. Limitations of monitoring methods

[0004] (1) Existing methods rely on single time or short time series images. Many current remote sensing monitoring methods rely on single time images or only use short time series of image data. Early research mainly relies on single or a small number of optical images. Although this approach can provide more intuitive information in some cases, it lacks continuity in the time dimension and cannot fully reveal the dynamic change process of the ground object. Short time series data may not be able to reflect the impact of long-term factors such as seasonal changes and climate changes on the ecological environment and productivity of rubber plantations.

[0005] (2) Difficulty in fully utilizing historical remote sensing data. With the development of remote sensing technology, historical remote sensing image data has become increasingly abundant. Past image data contains a wealth of valuable information that can provide a basis for long-term trend analysis. However, existing monitoring methods often fail to fully exploit the potential of these historical data, resulting in insufficient understanding of historical changes.

[0006] (3) Insufficient exploration of the economic life cycle characteristics of rubber plantations. Existing methods do not fully explore the economic life cycle characteristics of rubber plantations.

[0007] The feature mining of the life cycle is still insufficient. The economic life cycle of a rubber plantation includes multiple stages, such as planting, seedling stage, growth and maturity stage, harvesting stage, etc. The characteristics of these stages are significantly different. Single image or short time series data cannot fully reveal the land use change and vegetation growth patterns of these different stages. In addition, the productivity of a rubber plantation is affected by multiple factors, such as climate change, management mode (such as fertilization, irrigation), pest control, etc. However, the existing methods are still insufficient in analyzing the interaction of these complex factors. Therefore, more diverse and in-depth remote sensing technology is needed to support the accurate monitoring and dynamic analysis of the economic life cycle of a rubber plantation.

[0008] These limitations make it difficult for existing technology to comprehensively and accurately monitor and evaluate the development status and environmental changes of a rubber plantation. Therefore, more diverse monitoring methods are needed, combining long time series, cross-source data, and high-resolution images to better extract the characteristics of each aspect of the life cycle of a rubber plantation.

[0009] 2. Incompleteness of feature extraction

[0010] (1) Failure to fully consider the characteristics of different growth stages of a rubber plantation. During the growth of a rubber plantation, the characteristics of different growth stages show significant differences. Each stage of a rubber tree from planting to maturity (such as seedling stage, juvenile stage, mature stage) has different growth characteristics and land feature change patterns. Existing remote sensing monitoring methods may not fully consider these differences between stages, leading to the neglect of dynamic changes between growth stages during feature extraction. For example, the coverage of rubber trees in the seedling stage is low, and the vegetation signal is weak, while in the mature stage, it has strong canopy reflection characteristics. If these different characteristics of different growth stages are not considered, it may lead to biased monitoring results and fail to accurately reflect the status of rubber plantations in different growth stages.

[0011] (2) Insufficient identification of surface disturbance characteristics in the early stage of plantation construction. In the early stage of rubber plantation construction, large-scale land leveling, vegetation removal, and other operations are usually carried out on the ground, which will leave obvious features in remote sensing images and affect the subsequent vegetation growth patterns. However, existing feature extraction methods may not fully identify the surface disturbance characteristics in the early stage of plantation construction. For example, bare soil after land reclamation, traces of vegetation removal, sparseness of vegetation in the early stage after reclamation, etc. These features are crucial for the growth monitoring of subsequent rubber plantations. If these early disturbance features are ignored, it may affect the analysis of subsequent growth trends and lead to insufficient accuracy in judging the construction stage of a rubber plantation.

[0012] (3) Insufficient utilization of phenological characteristics. Phenological characteristics of plants (such as leaf area index, greenness index, seasonal variation, etc.) are important indicators reflecting the growth status of vegetation and can help analyze the health status, productivity level, and response to environmental changes of vegetation growth. However, existing remote sensing monitoring methods may not fully utilize the phenological characteristics of rubber plantations. For example, the growth of rubber trees is affected by a variety of factors such as seasonal changes, climate conditions, fertilization, and irrigation. Phenological characteristics can reflect the combined effects of these factors on the growth of rubber plantations. If phenological characteristics are not fully utilized, in-depth analysis of vegetation growth patterns, yield prediction, and health diagnosis may be missed. In addition, the utilization of phenological characteristics usually relies on the support of long-term remote sensing data. By analyzing the seasonal changes of vegetation, key turning points in the growth cycle (such as leaf fall and leaf emergence) can be revealed. If the monitoring methods lack sufficient time-series data or ignore the changes in different phenological stages, it may lead to the inability to accurately capture important information in the growth cycle of rubber plantations, affecting the assessment of growth health status.

[0013] 3. Uncertainty regarding application effectiveness

[0014] (1) Poor accuracy under complex terrain conditions. Complex terrain (such as hills and mountains) can affect the acquisition and analysis of remote sensing images. In particular, when the terrain is undulating, the ground feature signals in the remote sensing images are easily affected by shadows and occlusions, thus reducing the monitoring accuracy. For rubber plantations, especially on mountains or slopes, the reflectivity of the tree canopy may deviate significantly due to factors such as terrain slope and solar angle. This terrain influence makes land cover classification and feature extraction based on remote sensing more difficult, especially during automated processing, which may result in misclassification and inaccurate analysis results. Therefore, under complex terrain conditions, the accuracy of remote sensing monitoring often fails to meet expectations, limiting the accurate assessment of rubber plantations.

[0015] (2) Limited ability to process mixed pixels. Pixels in remote sensing images typically represent a certain area of ​​the land surface, but due to resolution limitations, many pixels may contain multiple land cover types (e.g., rubber plantations and surrounding farmland, woodland, etc., are mixed in the same pixel). This phenomenon is called "mixed pixels." Currently, many remote sensing analysis methods have limited ability to process mixed pixels, resulting in lower accuracy in land cover classification and feature extraction under such circumstances. Other crops or natural vegetation may exist around the rubber plantation.

[0016] These mixed signals are difficult to classify into a single category, especially in lower-resolution images. Failure to effectively demix them (e.g., using high-resolution imagery, band-weighted methods, or machine learning techniques) leads to errors and uncertainties in the monitoring results.

[0017] (3) There is a large error in the estimation of the establishment year. When using remote sensing images to monitor rubber plantations, accurate estimation of the establishment year is very important, as it helps to analyze the growth stage, historical development, and production potential of the rubber plantation. However, due to the limitations of spatial and temporal resolution of remote sensing images, there is often a large error in the estimation of the establishment year. The establishment process of the rubber plantation may not be clearly visible in the image, especially in the early stage of plantation establishment, which may only have a small amount of vegetation or land disturbance, resulting in remote sensing data failing to clearly capture the time point of plantation establishment. In addition, the initial growth of rubber trees is usually slow and may not be accurately identified by conventional features such as vegetation index in remote sensing images. Therefore, the estimation of the establishment year often relies on other indirect methods (such as ground surveys, historical records, etc.), which may result in large errors due to data deficiency or inconsistency. Such errors not only affect the accuracy of the plantation time, but also may affect the subsequent growth trend and yield prediction. SUMMARY

[0018] Therefore, the present application provides a method for identifying the distribution and establishment time of rubber plantations based on full-time sequence images.

[0019] To solve the above technical problems, the present application adopts the following technical solutions:

[0020] The method for identifying the distribution and establishment time of rubber plantations based on full-time sequence images comprises the following steps:

[0021] Step 1, data acquisition and preprocessing:

[0022] Data acquisition includes: long-time sequence optical image data, high-resolution optical data, SAR data, auxiliary data;

[0023] Data preprocessing includes: optical data preprocessing, SAR data preprocessing, vegetation index calculation;

[0024] Step 2, time sequence analysis based on LandTrendr and time correction to determine the establishment time of the plantation: first, use the LandTrendr algorithm to analyze forest disturbance, then correct the time according to the growth characteristics of rubber, determine the establishment year of the plantation according to the establishment year determination standard;

[0025] Step 3, extraction of key features of economic life cycle: including juvenile period features, mature period features, spectral, phenological and structural features of mapping year;

[0026] Step 4, random forest classification and identification: create a classification data feature set and perform classification;

[0027] Step 4, random forest classification and identification: create a classification data feature set and perform classification;

[0028] Step 5, precision verification and evaluation.

[0029] Preferably, in step 1, the long-time optical image data includes all Landsat 4 / 5 / 7 / 8 / 9 Level 2 Collection 2 Tier 1 data provided by the United States Geological Survey;

[0030] The high-resolution optical data includes monthly and semi-annual Planet images spliced, with a spatial resolution of 4.77 meters and four spectral bands of blue, red, green and near-infrared;

[0031] The SAR data includes horizontal emission horizontal reception (HH) and horizontal emission vertical reception (HV) polarization bands, stored as 16-bit digital values;

[0032] The auxiliary data includes DEM elevation data, land cover data, and administrative boundary data.

[0033] Preferably, in step 1, the optical data preprocessing includes radiometric correction and geometric correction;

[0034] Radiometric correction is: Landsat and Sentinel-2 both convert the digital values of the sensor into radiometric brightness through the calibration coefficient, and then perform atmospheric correction to obtain the ground reflectivity;

[0035] Geometric correction is: image registration and projection correction are performed on Landsat and Sentinel-2 to ensure that the geographic coordinates of the remote sensing image are consistent with the actual ground.

[0036] Preferably, in step 1, the SAR data preprocessing includes orthorectification and slope correction;

[0037] The orthorectification is to correct the geometric distortion in the SAR image by using a digital elevation model or a digital surface model, ensuring that each pixel corresponds to the actual ground position; convert the SAR image from its original sensor coordinate system to the required map coordinate system, and adjust the spatial position;

[0038] The slope correction is to correct according to the slope, azimuth and incidence angle of the ground object using a theoretical model; by adjusting the reflection intensity of the image, the influence of slope is removed.

[0039] Preferably, in step 1, the vegetation index calculation refers to the calculation of the normalized difference vegetation index, the surface water index and the normalized burning index in land use change monitoring, and the calculation formula is as shown in (1),

[0040] (2), (3) are shown:

[0041] (1)

[0042] (2)

[0043] (3)

[0044] In the formula, P red , PNIR , PSWIR 1, and PSWIR 2 respectively represent the reflectivity of the red light band, the near-infrared band, the short-wave infrared band 1 and the short-wave infrared band 2 of the remote sensing image.

[0045] Preferably, in the step 2, the LandTrendr algorithm parameter optimization value setting range is: maxSegments: 8-15, spikeThreshold: 0.3-0.7, vertexCountOvershoot: 1-2, recoveryThreshold: 0.3-0.7, minObservationsNeeded: 3-5; preferably, the parameter optimization value is set as: maxSegments=11, spikeThreshold=0.5, vertexCountOvershoot=1, recoveryThreshold=0.5, minObservationsNeeded=4.

[0046] Preferably, in the step 2, the establishment year of the rubber plantation is determined as the year that meets the following four standards in reverse order:

[0047] Lower than the minimum NBR threshold value is TsNBR _ MIN ;

[0048] NBRSTART _ IRP The lowest NBR in the three-year buffer zone, with the starting vertex of the gain fitted by the LandTrendr algorithm as the reference;

[0049] Gain amplitude NBRMAG _ IRP Higher than the minimum amplitude threshold value Ts Δ NBR ;

[0050] The duration between the starting point and the ending point NDUR _ IRP Greater than two years.

[0051] Preferably, in said step 3, the young age characteristics refer to the period from the plantation establishment year to the mature year, in the unripe stage or unripe rubber plantation, the selection characteristics NBRSTART IRP NBRMAG IRP

[0052] NDUR IRP

[0053] The beginning year of the mature stage characteristics is determined by adding the plantation establishment year to the number of years usually required for maturation NDUR IRP LSWIMIN LSWISTD FLSWI <0. 15, corresponding to the season of rubber leaf senescence, defoliation and leafing, is called SDF season; the average values of these mature stage characteristics are calculated and denoted as

[0054] LSWIMIN AVG SDF MRP LSWISTD AVG SDF MRP FLSWI <0.15 AVG SDF MRP

[0055] The mapping year characteristics include optical characteristics and SAR characteristics.

[0056] Preferably, in said step 3, the optical characteristics include annual period statistical characteristics, growth period characteristics, leaf senescence, defoliation and leafing period characteristics, and leaf bud development period characteristics.

[0057] The annual period statistical characteristics are the annual median values of Sentinel-2 SWIR1, NIR and red light bands and NDVI, denoted as S 2 _ swir1 MED ANN S 2 _ nirMED ANN S 2_ redMED ANN S 2 _ NDVIMED _​​​​​​​​​​​​​​​​​​​​​​​​​​​​ANN ; for Planet imagery, the median of the blue, green, red and NIR bands and of the NDVI are computed, respectively denoted Pl blueMED ANN , Pl greenMED ANN , Pl redMED ANN , Pl nirMED ANN and Pl NDVIMED ANN ;

[0058] Growth period features are the computation of the mean NDVI, denoted NDVIAVG GRN ; the interval mean of NDVI and LSWI between the 0th and 10th quantiles, denoted NDVIIM 010 GRN and LSWIIM 010 GRN ; and the frequency of NDVI values lower than 0.6 in LS2, denoted FNDVI <0.6 GRN ;

[0059] Leaf senescence, defoliation and leaf flushing period features are the computation of the mean, minimum and standard deviation of LSWI using the mapping year and the previous year imagery, denoted LSWIAVG SDF , LSWIMIN SDF , LSWIMIN SDF ;

[0060] Leaf bud development period features are the computation of the median of the blue, green, red, rel, re2, NIR and SWIR1 bands of Sentinel-2.

[0061] SAR features are the computation of the annual median of the HH and HV polarizations of PALSAR-2 of the mapping year.

[0062] Preferably, in step 4, the specific method is to gather all features of the immature and mature rubber plantation stages, as well as the features of the annual period, the leaf green period, the leaf senescence, defoliation and leaf flushing period and the leaf bud development period of the mapping year, and the elevation and slope of the DEM data, to create a comprehensive feature set.

[0063] The present application achieves the following technical effects with respect to the prior art: ​​​​​​​​​​​​​​​​​​​​​

[0064] The application realizes accurate identification of the spatial distribution and establishment year of rubber plantations, and the method has strong adaptability and generalizability, and shows significant identification advantages under complex landscape conditions. BRIEF DESCRIPTION OF DRAWINGS

[0065] Fig. 1 is a schematic diagram of the rubber plantation distribution and establishment time identification method based on full-time sequence images according to the application;

[0066] Fig. 2 is a LandTrendr fitting sequence diagram in the rubber plantation distribution and establishment time identification method based on full-time sequence images according to the application;

[0067] a) ideal LandTrendr fitting; b) corrected LandTrendr fitting; c) backward direction search starting point; d) forward direction search starting point; e) super long gain that needs to move the starting point forward; f) extreme fitting with only one gain;

[0068] Fig. 3 is a Vietnam rubber distribution and establishment time precision evaluation scatter plot of the rubber plantation distribution and establishment time identification method based on full-time sequence images according to the application. DETAILED DESCRIPTION

[0069] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0070] The application discloses a rubber plantation distribution and establishment time identification method based on full-time sequence images, including the following steps:

[0071] Step 1, data acquisition and pretreatment:

[0072] Data acquisition includes: long-time sequence optical image data, high-resolution optical data, SAR data, auxiliary data;

[0073] Data pretreatment includes: optical data pretreatment, SAR data pretreatment, vegetation index calculation;

[0074] Long-time sequence optical image data includes: all Landsat 4 / 5 / 7 / 8 / 9 Level 2 Collection 2 Tier 1 data provided by the United States Geological Survey;

[0075] High resolution optical data includes monthly and semi-annual mosaics with a spatial resolution of 4.77 meters and four spectral bands: blue, red, green, near infrared;

[0076] SAR data includes horizontal transmit horizontal receive (HH) and horizontal transmit vertical receive (HV) polarized bands, stored as 16-bit digital numbers;

[0077] Auxiliary data includes DEM elevation data, land cover data, administrative boundary data;

[0078] Optical data pre-processing includes radiometric correction and geometric correction;

[0079] Radiometric correction is: Landsat and Sentinel-2 both convert the sensor's digital numbers to radiance through a scaling coefficient, and then perform atmospheric correction to obtain the ground reflectance;

[0080] Geometric correction is: image registration and projection correction are performed on Landsat and Sentinel-2 to ensure that the geographic coordinates of remote sensing images are consistent with the actual ground;

[0081] SAR data pre-processing includes ortho-rectification and slope correction;

[0082] The ortho-rectification is to correct the geometric distortion in the SAR image by using a digital elevation model or a digital surface model, to ensure that each pixel corresponds to the actual ground position; convert the SAR image from its original sensor coordinate system to the required map coordinate system, and adjust the spatial position;

[0083] The slope correction is to use a theoretical model to correct according to the slope, azimuth and incidence angle of the ground object; by adjusting the reflection intensity of the image, the influence of slope is removed;

[0084] Vegetation index calculation refers to the calculation of normalized difference vegetation index, surface water index and normalized burn index in land use change monitoring, and the calculation formulas are shown in (1), (2) and (3):

[0085] (1)

[0086] (2)

[0087] (3)

[0088] In the formula, P red , PNIR , PSWIR 1, and PSWIR2 represent the reflectance of red band, near-infrared band, short-wave infrared band 1 and short-wave infrared band 2 of remote sensing image respectively;

[0089] Step 2, LandTrendr-based time series analysis: first, LandTrendr algorithm parameter optimization is performed, and then the year determination standard is established;

[0090] Preferably, the parameter optimization value is set as: maxSegments=11, spikeThreshold=0.5, vertexCountOvershoot=1, recoveryThreshold=0.5, and minObservationsNeeded=4.

[0091] The establishment year of the rubber plantation is determined as the year that meets the following four standards in reverse order:

[0092] Lower than the minimum NBR threshold value is TsNBR MIN ;

[0093] NBRSTART IRP The minimum NBR in the three-year buffer zone, with the starting vertex of the gain fitted by the LandTrendr algorithm as the reference;

[0094] Gain amplitude NBRMAG IRP Higher than the minimum amplitude threshold value Ts Δ NBR ;

[0095] The duration between the starting point and the ending point NDUR IRP Greater than two years;

[0096] Step 3, economic life cycle feature extraction: including juvenile period feature, mature period feature, spectral, phenological and structural features of the mapping year;

[0097] Among them, the juvenile period feature refers to the feature selected in the immature stage or immature rubber plantation from the establishment year of the rubber plantation to the mature year NBRSTART IRP , NBRMAG IRP ​​​​​​and NDUR IRP classification is performed;

[0098] The starting year of the maturation phase characteristics is determined by adding the plantation establishment year to the number of years usually required for maturation NDUR IRP Each maturation year is generated using imagery acquired between day 1 and day 100 of the year LSWIMIN , LSWISTD and the frequency of LSWI less than 0.15 FLSWI <0.15, corresponding to the season of rubber leaf senescence, defoliation and leaf fall, is called SDF season; the average of these characteristics in the maturation phase is calculated and denoted respectively by

[0099] LSWIMIN AVG SDF MRP , LSWISTD AVG SDF MRP and FLSWI <0.15 AVG SDF MRP ;

[0100] The mapping year characteristics include optical and SAR characteristics;

[0101] The optical characteristics include annual period statistics, growth phase characteristics leaf senescence, defoliation and leaf emergence, and leaf bud development phase characteristics;

[0102] wherein the annual period statistics are the median values of the SWIR1, NIR and red bands of Sentinel-2 and NDVI, denoted respectively by S 2_swir1 MED ANN , S 2 nirMED ANN , S 2 redMED ANN and S 2 NDVIMED ANN ; for Planet imagery, the median values of the blue, green, red and NIR bands and NDVI are calculated and denoted respectively by Pl blueMED ANN , Pl greenMED ANN ,​​​​​​​​​​​​​​​​​​​​​​​Pl redMED ANN Pl nirMED ANN Pl NDVIMED ANN

[0103] The growth period features are the average NDVI, denoted as NDVIAVG GRN ; the average of the NDVI and LSWI interval between the 0th and 10th quantiles, denoted as NDVIIM GRN LSWIIM GRN ; and the frequency of NDVI values less than 0.6 in LS2, denoted as FNDVI GRN

[0104] The leaf senescence, defoliation and leaf flushing period features are the average, minimum and standard deviation of the LSWI computed using the mapping year and the previous year imagery, denoted as LSWIAVG SDF LSWIMIN SDF LSWIMIN SDF

[0105] The leaf bud development period features are the median of the Sentinel-2 blue, green, red, rel, re2, NIR and SWIR1

[0106]

[0107] The SAR features are the annual median of the HH and HV polarizations of the PALSAR-2 in the mapping year;

[0108] Step 4, Random Forest classification identification:

[0109] All features of the immature and mature rubber plantation stages are summarized, as well as the features of the annual period, the leaf green period, the leaf senescence, defoliation and leaf flushing period and the leaf bud development and growth period of the mapping year, and the elevation and slope of the DEM data, to create a comprehensive feature set;

[0110] Step 5, Precision verification and evaluation.

[0111] Example 1:

[0112] As Figure 2 ​​​​​​​​​​​​​​​​​​​​​​As shown, there are six typical LandTrendr fitting sequences during the recognition process. Ideally, the starting point should closely correspond to the starting vertex of the last gain fitted by the LandTrendr algorithm. Figure 2 a) In the second case, the LandTrendr algorithm identifies several consecutive short gains, each lasting less than six years, representing short-term fluctuations. These consecutive short gains should be combined to determine the conformity. Ts Δ NBR The appropriate endpoints for the standard (Fig. 2b). The starting year is determined by the minimum NBR value found around the latest gain starting vertex within the three-year buffer (Fig. 2b). The minimum NBR value near the gain starting vertex can be searched in both backward (Fig. 2c) and forward (Fig. 2d) directions. Another case involves ultra-long gains that meet all criteria, but the establishment year is determined too early due to the excessively long immature phase in rubber plantations (Fig. 2e). The immature phase typically lasts about 7 years, but can be up to 10 years in marginal areas. Considering that the LT fitting period may contain more points, gains exceeding 12 years are defined as ultra-long gains. In this case, the first of the latest long gains that meets the criteria is determined. NBRSTART _ IRP < TsNBR _ MIN The reverse point is then used to search for the minimum NBR value forward with a 3-year buffer. If this minimum NBR value is used as the starting point and all criteria are met, the year is updated to the year of the minimum NBR found. Figure 2 e). An extreme case arises when there is only one gain in the entire series (Figure 2f). With the first gain less than... TsNBR _ MIN Using the reverse point as a reference, determine the minimum NBR value within a 3-year buffer. This point will serve as the starting point, and a forward search will be performed with a 12-year buffer (the threshold for ultra-long gain) to find the highest NBR.

[0113] As the endpoint. If both points meet all five criteria, the starting point year is set to the year the plantation was established. If no gain meets four criteria (a special case), the plantation establishment year is set to 1986—the year when most areas began to have multiple Landsat TM images. TsNBR _ MIN and Ts Δ NBR The values ​​were determined by analyzing the distribution of ground samples and were set to 520 and 150, respectively, rounded to the nearest whole number.

[0114] Threshold determination in the recognition process: NBRmin can be adjusted in the range of 400-600, and Magmin can be adjusted in the range of 100-200.

[0115] Example 2: Spatial distribution accuracy verification and evaluation

[0116] The rubber map generated by the method is evaluated by the confusion matrix composed of 40% randomly selected ground samples, which are separated from the original data set to ensure representativeness and unbiased verification; the classification achieves high accuracy in all indicators, with producer accuracy (PA) of 96.92%, user accuracy (UA) of 89.87%, and F1 score of 0.93; the overall accuracy (OA) reaches 93.75%, and the kappa coefficient is 0.87, indicating high consistency with the ground truth data.

[0117] Non-rubber classification performs better, with UA of 97.35%, while rubber is 89.87%. The error of rubber and non-rubber is 10.13% and 2.65% respectively, while the omission error is 3.08% for rubber and 8.80% for non-rubber.

[0118] Significantly better than existing methods (e.g. 75.00% accuracy of WangR_2021).

[0119] Example 3: Establishment year accuracy verification and evaluation

[0120] The establishment year data of the algorithm is evaluated using 100 random samples generated on the classified rubber distribution map by the "ee.Image.sample" method in the GEE platform. The observed values come from the historical images of Google Earth to intuitively explain the establishment year of rubber plantations. As shown in Figure 3, the observed years and estimated establishment years of rubber plantations are compared, and the linear fitting is calculated, with the determination coefficient (R²) : 0.99, and the root mean square error: 0.25 years.

[0121] Obviously, it is highly consistent with the actual statistical data.

[0122] The above is only the preferred embodiment of the present application, and does not limit the technical scope of the present application, so any slight modification, equivalent change and modification of the above embodiment according to the technical essence of the present application still belongs to the scope of the technical solution of the present application.

Claims

1. A method for identifying the distribution and the establishment time of a rubber plantation based on full-time sequence images, characterized in that, The method comprises the following steps: Step 1, data acquisition and preprocessing: Data acquisition includes long time series optical image data, high resolution optical data, SAR data, auxiliary data; Data preprocessing includes optical data preprocessing, SAR data preprocessing, vegetation index calculation; Step 2, time series analysis based on LandTrendr and time correction to determine the establishment year of the plantation: first, the forest disturbance analysis is carried out by using the LandTrendr algorithm, then the time correction is carried out according to the growth characteristics of rubber, the establishment year is determined according to the establishment year determination standard and the establishment year of the plantation is determined; Step 3, extraction of key characteristics of economic life cycle: including juvenile period characteristics, mature period characteristics, mapping year phenology, spectrum and structure characteristics; Step 4, random forest classification and identification of rubber plantation: creating a classification data feature set and executing classification; Step 5, precision verification and evaluation; In the step 1, the vegetation index calculation refers to calculating the normalized difference vegetation index NDVI, the land surface water index LSWI and the normalized burn index NBR respectively, and the calculation formulas are as follows: wherein, , , , and respectively represent reflectance of red, near-infrared, short-wave infrared band 1, and short-wave infrared band 2 of the remote sensing image. In the step 2, the establishment year of the rubber plantation is determined as the year that meets the following four standards in reverse order: (1) the NBR for the year is lower than a set minimum NBR threshold ; (2) is the lowest NBR within the three-year buffer, referenced to the starting vertex of the gain fitted by the LandTrendr algorithm; (3) gain magnitude above the minimum magnitude threshold ; (4) duration between start and end points more than two years; In step 3, the young age characteristics refer to the characteristics selected in the unripe stage or unripe rubber plantations, from the establishment year to the mature year in the rubber garden , and classification; Using imagery acquired between day 1 and 100 of the year to generate each of the mature years , and LSWI, the frequency of LSWI being less than 0.15; The mapping year characteristics include optical characteristics and SAR characteristics.

2. The all-time-lapse video based rubber plantation distribution and establishment time identification method according to claim 1, characterized in that, In the step 1, The high resolution optical data includes monthly and semi-annual Planet images with a spatial resolution of 4.77 meters and four spectral bands of blue, red, green and near infrared; The SAR data includes PALSAR-2 horizontal transmission horizontal reception and horizontal transmission vertical reception polarization wave bands, stored as 16-bit digital values; The auxiliary data includes DEM elevation data, land cover data and administrative boundary data.

3. The all-time-lapse video based rubber plantation distribution and establishment time identification method according to claim 1, characterized in that, In the step 1, the optical data preprocessing includes radiation correction and geometric correction; Radiometric correction is that Landsat and Sentinel-2 both convert the digital values of the sensor into radiance through a scaling coefficient, and then perform atmospheric correction to obtain the ground reflectance; Geometric correction is that image registration and projection correction are performed on Landsat and Sentinel-2 to ensure that the geographic coordinates of remote sensing images are consistent with the actual ground.

4. The all-time-lapse video based rubber plantation distribution and establishment time identification method according to claim 1, characterized in that, In the step 1, the SAR data preprocessing includes orthorectification and slope correction; The orthorectification is to correct the geometric distortion in the SAR image by using a digital elevation model or a digital surface model, so that each pixel is consistent with the actual ground position; the SAR image is converted from its original sensor coordinate system to the required map coordinate system, and the spatial position is adjusted; The slope correction is to correct according to the slope, azimuth and incidence angle of the ground object by using a theoretical model; by adjusting the reflection intensity of the image, the influence of slope is removed.

5. The all-time-lapse video based rubber plantation distribution and establishment time identification method according to claim 1, characterized in that, In step 2, the LandTrendr algorithm parameter optimization value setting range is: maxSegments: 8-15, spikeThreshold: 0.3-0.7, vertexCountOvershoot: 1-2, recoveryThreshold: 0.3-0.7, minObservationsNeeded: 3-5.

6. The all-time-lapse video based rubber plantation distribution and establishment time identification method according to claim 1, wherein, In step 4, the specific method is: all features of the immature rubber plantation and the mature rubber plantation stage are summarized, and the features of the annual period, the chlorophyll period, the leaf aging, the defoliation and the leaf budding and development period of the mapping year, and the elevation and slope of the DEM data are summarized to create a comprehensive feature set.

Images