CN117911783A - Young rubber plantation identification method based on multi-source remote sensing - Google Patents
Young rubber plantation identification method based on multi-source remote sensing Download PDFInfo
- Publication number
- CN117911783A CN117911783A CN202410134404.3A CN202410134404A CN117911783A CN 117911783 A CN117911783 A CN 117911783A CN 202410134404 A CN202410134404 A CN 202410134404A CN 117911783 A CN117911783 A CN 117911783A
- Authority
- CN
- China
- Prior art keywords
- rubber
- young
- planting
- remote sensing
- plantation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000007637 random forest analysis Methods 0.000 claims abstract description 30
- 238000010586 diagram Methods 0.000 claims abstract description 17
- 230000008859 change Effects 0.000 claims abstract description 9
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 9
- 230000012010 growth Effects 0.000 claims abstract description 8
- 238000012876 topography Methods 0.000 claims abstract description 8
- 239000002689 soil Substances 0.000 claims description 8
- 230000003595 spectral effect Effects 0.000 claims description 8
- 238000012544 monitoring process Methods 0.000 claims description 7
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 7
- 230000015572 biosynthetic process Effects 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000003786 synthesis reaction Methods 0.000 claims description 6
- 244000043261 Hevea brasiliensis Species 0.000 claims description 4
- 238000003066 decision tree Methods 0.000 claims description 4
- 101000686491 Platymeris rhadamanthus Venom redulysin 1 Proteins 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000005855 radiation Effects 0.000 claims description 3
- 238000001308 synthesis method Methods 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 claims description 2
- 238000002310 reflectometry Methods 0.000 claims description 2
- 238000001228 spectrum Methods 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 2
- 230000006872 improvement Effects 0.000 description 9
- 239000003638 chemical reducing agent Substances 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000873 masking effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 101710179734 6,7-dimethyl-8-ribityllumazine synthase 2 Proteins 0.000 description 1
- 241000196324 Embryophyta Species 0.000 description 1
- 101710186609 Lipoyl synthase 2 Proteins 0.000 description 1
- 101710122908 Lipoyl synthase 2, chloroplastic Proteins 0.000 description 1
- 101710101072 Lipoyl synthase 2, mitochondrial Proteins 0.000 description 1
- 206010035148 Plague Diseases 0.000 description 1
- 241000607479 Yersinia pestis Species 0.000 description 1
- 230000003698 anagen phase Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 239000002352 surface water Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000002054 transplantation Methods 0.000 description 1
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to a young rubber plantation identification method based on multi-source remote sensing, which aims to improve classification accuracy. The method comprises the following steps: determining potential planting areas of rubber; according to the potential planting area of the rubber, combining the growth characteristics of the rubber seedlings, and synthesizing a time sequence data set with a certain step length by using a multi-source satellite remote sensing image source; and classifying by using a random forest model according to the synthesized time sequence data set to obtain a distribution diagram of the young rubber plantation. The invention relates to young rubber plantation space distribution information extraction, which comprises the steps of firstly determining potential planting areas of rubber, secondly combining growth characteristics of rubber seedlings, synthesizing a time sequence data set with a certain step length by using a satellite remote sensing image source, and finally classifying by using a random forest model to obtain a young rubber plantation distribution map. The method improves the classification precision of young rubber plantation by comprehensively considering the factors such as the topography condition, land utilization change, rubber planting history distribution frequency and the like.
Description
Technical Field
The invention relates to the technical field of agricultural remote sensing, in particular to a young rubber plantation identification method based on multi-source remote sensing.
Background
The accurate identification of young rubber plantation (< 6 years) is an important scientific problem in remote sensing research of rubber plantation, and the problem is not only directly related to the accuracy of monitoring indexes such as area and distribution of the rubber plantation, but also influences the practical application results such as growth monitoring, daily operation management and yield estimation of the rubber plantation. However, because the crown canopy of young rubber plantations is low in canopy closure, and does not have the crown texture typical of mature rubber plantations, existing algorithms suitable for identifying mature rubber plantations cannot be directly applied to the identification of young rubber plantations. In addition, the interference of soil types and other vegetation, such as shrubs, also causes young rubber plantations to exhibit mixed spectral reflectance characteristics, which spectral mixing between land use types further increases the difficulty of accurately identifying young rubber plantations. Therefore, how to realize effective identification of young rubber plantation is always a great technical problem which plagues monitoring of rubber plantation industry. The past research has focused mainly on mature rubber plantations with obvious forest characteristics, and there is a large error in the identification of young rubber plantations, which directly results in the current acquired rubber planting profile being significantly delayed from the actual planting situation by 5-6 years. Therefore, developing a young rubber plantation identification method based on multi-source remote sensing images has important practical significance for solving the problems.
Despite the challenges described above, young rubber plantations still exhibit certain spectral and climatic characteristics. In the initial stage of planting, the surface soil around the rubber seedlings can be exposed, and the characteristics can be effectively captured by using the surface water index LSWI, so that the potential rubber planting area can be determined. In addition, early rubber seedlings in the rapid growth phase experience frequent morphological and spectral changes. Thus, constructing a time series of cloudless weather images containing appropriate time intervals is critical to capturing these dynamically changing features.
Disclosure of Invention
The invention aims to provide a young rubber plantation identification method based on multi-source remote sensing, which has the advantages of high calculation efficiency and high prediction precision.
The invention aims to solve the problems in the prior art.
Compared with the prior art, the technical scheme of the invention has the following beneficial effects:
A young rubber plantation identification method based on multi-source remote sensing comprises the following steps: determining potential planting areas of rubber; according to the potential planting area of the rubber, combining the growth characteristics of the rubber seedlings, and synthesizing a time sequence data set with a certain step length by using a multi-source satellite remote sensing image source; and classifying by using a random forest model according to the synthesized time sequence data set to obtain a young rubber plantation distribution map.
As a further improvement, the determining the potential planting area of the rubber includes: defining a potential rubber planting area according to the topography condition of rubber planting; rubber plantations are built in areas with relatively flat topography; and acquiring elevation data of the target area by using SRTMGL1_003 images of the USGS, calculating gradient by using an ee.terrain function in a GEE platform, and setting a threshold value to exclude areas in which the elevation and the gradient are not consistent.
As a further improvement, the determining the potential planting area of the rubber further comprises: defining a potential rubber planting area according to the area excluding unsuitable rubber tree planting; using google near real-time land cover data Dynamic World to exclude areas classified as buildings and bodies of water; dynamic World synthetic images of the current year were monitored using the Dynamic World function of geemap packets in python, and the label band of the Dynamic World synthetic images was used to exclude water (label=0) and building (label=6) areas.
As a further improvement, the determining the potential planting area of the rubber further comprises: defining a potential rubber planting area according to the land utilization change during the establishment of the plantation; monitoring bare soil by using LSWI, and screening a rubber potential planting area with land utilization change according to the condition that the minimum value of the LSWI is less than 0.1; screening out remote sensing images of 4-12 months in Landsat 8 and Sentinel-2 image sets of corresponding years, and obtaining an LSWI minimum distribution map in the whole target area by using minimum synthesis, namely calling a reduction (ee. Reduction. Min ()) function in a GEE platform, wherein the area with the LSWI minimum value larger than 0.1 is excluded.
As a further improvement, the determining the potential planting area of the rubber further comprises: potential rubber planting areas are defined according to the historical distribution frequency of rubber planting; obtaining a rubber planting distribution frequency chart by superposing a rubber plantation distribution chart with a target area of 30 years and a step length of 5 years; and expanding the area of each rubber planting distribution diagram by using a neighborhood maximum filtering mode, and summing all images to obtain a final rubber planting distribution frequency diagram, so that the area with the planting frequency of 0 is eliminated.
As a further improvement, the synthesizing a time sequence data set with a certain step length by using a multi-source satellite remote sensing image source according to the potential planting area of rubber and combining the growth characteristics of rubber seedlings comprises the following steps: land resource satellite Landsat 8/9, sentencel-2 multispectral image of the Sentinel No. 2 and synthetic aperture radar image of sentencel-1 of the Sentinel No. 1 are used in GEE; landsat 8/9 and Sentinel-2TOA optical images have been subjected to radiation and geometry correction using the atmospheric top reflectance TOA product; for the Sentinel-2 image, cloud score+ is used to control the Cloud and shadows in the band mask image in the data product; for Landsat 8/9 images, utilizing clouds and shadows in QA_PIXEL band and QA_ RADSAT band mask images; time series of 3 month, 6 month and 12 month steps were synthesized using the median synthesis method, respectively.
As a further improvement, for Landsat 8/9 and Sentinel-2 images, NDVI and LSWI vegetation indices for each time period were calculated using formulas (1) and (2), respectively:
Wherein ρ Red,ρNIR and ρ SWIR1 are red, near infrared and short wave infrared 1 bands of Landsat 8/9 and Sentinel-2, respectively.
As a further improvement, for Sentinel-2 images, the additional REP index needs to be calculated using equation (3):
Wherein ρ Red1,ρred2 and ρ red3 are red side band 1, red side band 2 and red side band 3 of Sentinel-2 respectively, and are corresponding to B5, B6 and B7 bands of Sentinel-2; the original spectrum band and the vegetation index band of the time sequence are used as characteristic variables of classification and are input into a classification model.
As a further improvement, the classifying using the random forest model to obtain the young rubber plantation distribution map comprises: the random forest model can be called by using an ee.classifiier.smileRandomForest function in the GEE platform; setting the number of decision trees to 100, the remaining settings following default settings in the GEE platform; taking the characteristics of the time sequence as input, wherein the shared independent variables of Landsat 8/9 and Sentinel-2 comprise 5 spectral bands and 2 vegetation indexes; if the random forest model contains Sentinel-2 data, increasing REP vegetation index as an independent variable; for a random forest model containing Senitnel-1SAR data, increasing the VH and VV polarized radar backscattering coefficients by the independent variable set; the dependent variables of each random forest model are binary classifiers to distinguish young rubber gardens (labeled 1) from non-young rubber gardens (labeled 0).
As a further improvement, the classification using the random forest model to obtain the young rubber plantation distribution diagram further comprises: and inputting all wave bands of each image in the image set synthesized by 3 month step length as a model, predicting a classification label corresponding to each pixel, wherein all pixels with classification labels of 1 are the final distribution area of young rubber.
The beneficial effects of the invention are as follows:
The invention relates to extraction of spatial distribution information of rubber forests, which comprises the steps of firstly determining a potential planting area of young rubber, secondly synthesizing a time sequence with a certain step length by using a satellite remote sensing data source, and finally classifying by using a random forest model to obtain a young rubber distribution map; the classification precision of young rubber plantation is improved.
Drawings
Fig. 1 is a schematic diagram of a young rubber plantation identification method based on multi-source remote sensing.
Fig. 2 is a schematic diagram of a rubber early recognition algorithm according to an embodiment of the present invention.
FIG. 3 is a schematic diagram of determining potential rubber planting areas provided by an embodiment of the present invention.
Fig. 4 is a schematic diagram of 2015-2022 young rubber distribution provided by an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. Thus, the following detailed description of the embodiments of the invention, as presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
In the description of the present invention, the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the present invention, the meaning of "a plurality" is two or more, unless explicitly defined otherwise.
Referring to fig. 1 to 4, a young rubber plantation identification method based on multi-source remote sensing includes: determining potential planting areas of rubber; according to the potential planting area of the rubber, combining the growth characteristics of the rubber seedlings, and synthesizing a time sequence data set with a certain step length by using a multi-source satellite remote sensing image source; and classifying by using a random forest model according to the synthesized time sequence data set to obtain a young rubber plantation distribution map.
FIG. 2 (a) is a plot of land cover change at the initial stage of rubber planting; FIG. 2 (b) is a graph showing the initial vegetation index change in rubber planting; FIG. 2 (c) is a histogram of LSWI minimum for year of rubber planting.
Determining potential planting areas of young rubber: identifying potential planted areas of young rubber helps to remove irrelevant areas, thereby reducing classification errors. Several factors, such as the topography of the rubber planting, the areas where it is not possible to plant the rubber tree, and the variation in land utilization during plantation establishment, help define potential rubber planting areas.
The determining a potential planting area of rubber comprises:
defining a potential rubber planting area according to the topography condition of rubber planting;
rubber plantations are built in areas with relatively flat topography; for example, the sea island of the second large planting area in China has the altitude of most rubber gardens lower than 650m and the gradient lower than 30 degrees;
Acquiring altitude data of a target area by using SRTMGL1_003 images of the USGS, calculating gradient by using an ee.terrain function in a GEE platform, and excluding areas with the altitude higher than 650m and the gradient higher than 30 degrees; setting the threshold excludes areas where altitude and grade are not met.
The determining the potential planting area of the rubber further comprises:
defining a potential rubber planting area according to the area excluding unsuitable rubber tree planting;
using google near real-time land cover data Dynamic World to exclude areas classified as buildings and bodies of water;
Dynamic World synthetic images of the current year were monitored using the Dynamic World function of geemap packets in python, and the label band of the Dynamic World synthetic images was used to exclude water (label=0) and building (label=6) areas.
The determining the potential planting area of young rubber further comprises:
Defining a potential rubber planting area according to the land utilization change during the establishment of the plantation;
As shown in fig. 2 (a), the need to clean the surface soil for rubber seedling transplantation generally results in large-area bare surface soil, LSWI is sensitive to monitoring bare soil, and the minimum value of LSWI is a histogram of young rubber garden according to fig. 2 (c);
Monitoring bare soil by using LSWI, and screening a rubber potential planting area with land utilization change according to the condition that the minimum value of the LSWI is less than 0.1;
And screening out images of 4-12 months in Landsat 8 image sets of corresponding years, and obtaining an LSWI minimum value distribution map in the whole target area by using a minimum value synthesis mode, namely calling a reduction (ee. Reducer. Min ()) function in a GEE platform, wherein the area with the LSWI minimum value larger than 0.1 is excluded.
The planting of rubber gardens has strong historical continuity, and a newly planted rubber garden is usually positioned near an existing rubber garden, so that if the rubber garden has an age-forming rubber distribution diagram for many years, the area where rubber is planted once and at present can be obtained.
FIG. 3 (a) is a graph showing the frequency of planting rubber in Hainan island from 1990 to 2020; FIG. 3 (b) is a schematic view of a potential rubber planting area.
The determining the potential planting area of young rubber further comprises:
Potential rubber planting areas are defined according to the rubber planting history distribution frequency of the plantation;
Obtaining a rubber distribution frequency chart by superposing a rubber distribution chart of a step length of 5 years in the past 30 years of a target area; the higher the data value, the greater the probability of continuing to grow rubber.
And (3) expanding the area by using a neighborhood maximum filtering (reduceNeighborhood functions are used, the reducer is set to ee.reducer.max (), the kernel size is set to 1) for each rubber planting distribution diagram, and then summing all images to obtain a final rubber planting distribution frequency diagram, so that the area with the planting frequency of 0 is excluded.
The synthesizing a time sequence with a certain step length by using a satellite remote sensing data source comprises the following steps:
Using land resource satellite Landsat 8/9, sentencel-2 multispectral image of the sentencel number 2 and sentencel-1 synthetic aperture radar image of the sentencel number 1 in GEE, and using the atmospheric top reflectivity TOA product, carrying out radiation and geometric correction treatment on the Landsat 8/9 and sentencel-2 TOA images archived in the GEE platform (the basic information is shown in table 1);
For the Sentinel-2 image, cloud score+ is used to control the Cloud and shadows in the band mask image in the data product;
For Landsat 8/9 images, utilizing QA_PIXEL wave bands and QA_ RADSAT wave bands to mask and exclude cloudy and shadow observation Points (PIXELs);
basic information of remote sensing image obtained in Table 1
The time series of 3 month, 6 month and 12 month steps were synthesized using the median synthesis method (image synthesis using the reduction (ee. Reducer. Media ()) function within each time window), respectively.
For Landsat 8/9 and Sentinel-2 images, the NDVI and LSWI vegetation indices in each time period were calculated using formulas (1) and (2), respectively:
Wherein ρ Red,ρNIR and ρ SWIR1 are red, near infrared and short wave infrared 1 bands of Landsat 8/9 and Sentinel-2, respectively, and are B4, B5 and B6 bands of Landsat 8, and B4, B8 and B11 bands of Sentinel-2, respectively.
For Sentinel-2 images, the additional REP index needs to be calculated using equation (3):
Wherein ρ Red1,ρred2 and ρ red3 are red side band 1, red side band 2 and red side band 3 of Sentinel-2, respectively, and are B5, B6 and B7 bands of Sentinel-2.
The original spectral band and vegetation index band of the time series will be input into the classification model as features for classification.
The random forest model is a set of decision trees, and the excellent classification performance and the characteristic of difficult overfitting enable the random forest model to be widely applied to the field of crop identification.
The classification by using the random forest model to obtain the young rubber plantation distribution map comprises the following steps:
The random forest model can be called by using an ee.classifiier.smileRandomForest function in the GEE platform; considering the trade-off of operation efficiency and accuracy, the number of decision trees is set to 100, and the rest of the settings follow the default settings in the GEE platform;
Taking as input the characteristics of the time series, the shared arguments of Landsat 8/9 and Sentinel-2 include 5 spectral bands (blue, green, red, near-infrared and short-wave infrared 2) and two vegetation indices (NDVI, LSWI);
if the random forest model contains Sentinel-2 data, increasing a vegetation index REP as an independent variable;
For a random forest model containing Senitnel-1SAR data, the independent variable set increases the VH (vertical transmit and horizontal receive) and VV (vertical transmit and vertical receive) polarized radar backscatter coefficients;
The dependent variables of all random forest models are binary classifiers that distinguish young rubber from non-young rubber (young rubber labeled 1, non-young rubber labeled 0).
The random forest model tests different data source synthesis strategies (Landsat 8 alone, sentinel-2 alone, landsat 8 and Sentinel-2 fusion data (LS 2) and LS2+Sentinel-1 combined) and synthesis strategies of different time steps (3 months, 6 months and 12 months), and finally determines that LS2+Sentinel-1 combined is used, and the random forest model matched with the 3 month time step has optimal classification performance, so that the model is adopted for inverting young rubber gardens.
The method for classifying by using the random forest model to obtain the young rubber plantation distribution map further comprises the following steps:
And (3) taking all wave bands (blue light, green light, red light, near infrared and short wave infrared 2 and NDVI, LSWI, REP, VH, VV) of each image in the image set synthesized in 3 month step length as model input, predicting classification labels (young rubber and non-young rubber) corresponding to each pixel, wherein all pixels with classification labels of 1 are final distribution areas of the young rubber.
Experimental conditions and content
1) The experimental site is selected from Hainan province of China, landsat 8TOA, sentinel-2TOA and Sentinel-1SAR satellite data of 2015-2022 are obtained in GEE, and necessary basic image preprocessing such as cloud mask and vegetation index calculation is completed.
2) Using the conditions of LSWI minimum <0.1, altitude <650m, slope <30 ° in the year, and excluding areas with Dynamic World classification as water and building, and rubber planting frequency of 0 in the last 30 years, the potential areas for rubber planting of each year 2015-2022 were obtained respectively.
3) Using young rubber gardens and non-rubber-like points planted in 2015 as training-like points, LS2+ Sentinel-1 time series data were constructed for a total of 5 years from 2015-2019 in 3 month steps. And taking all wave bands of the time sequence as input, taking the category of the training sample point as output, and establishing a random forest model for classifying the young rubber garden.
4) And classifying the 2015-2019 time series of the whole island of the Hainan island by using a random forest model, and then masking a 2015 rubber planting potential area to obtain a 2015 young rubber garden distribution map. Meanwhile, the models are respectively migrated to time sequences of 2016-2020, 2017-2021, 2018-2022, 2019-2022, 2020-2022, 2021-2022 and 2022 all year, and then masking is respectively carried out on the rubber planting potential areas corresponding to the initial years, so that a young rubber garden distribution diagram of each year of 2016-2022 is finally obtained, and the drawing precision is shown in table 2.
TABLE 2 evaluation of drawing precision of young rubber
The schematic diagram of the experimental effect is shown in fig. 4. FIG. 4 (a) is a 2015-2022 young rubber profile; FIG. 4 (b) uses classification results corresponding to time series of different lengths; FIG. 4 (c) is a comparison of the planting areas of young rubber in various city and counties in Hainan; fig. 4 (d) is a schematic diagram of total area of young rubber planted in 2015-2022 of hainan island.
The above examples are only for illustrating the technical scheme of the present invention and are not limiting. It will be understood by those skilled in the art that any modifications and equivalents that do not depart from the spirit and scope of the invention are intended to be within the scope of the appended claims.
Claims (10)
1. The young rubber plantation identification method based on multi-source remote sensing is characterized by comprising the following steps of:
determining potential planting areas of rubber;
according to the potential planting area of the rubber, combining the growth characteristics of the rubber seedlings, and synthesizing a time sequence data set with a certain step length by using a multi-source satellite remote sensing image source;
And classifying by using a random forest model according to the synthesized time sequence data set to obtain a young rubber plantation distribution map.
2. The method for identifying young rubber plantation based on multi-source remote sensing according to claim 1, wherein said determining potential planting area of rubber comprises:
defining a potential rubber planting area according to the topography condition of rubber planting;
rubber plantations are built in areas with relatively flat topography;
And acquiring elevation data of the target area by using SRTMGL1_003 images of the USGS, calculating gradient by using an ee.terrain function in a GEE platform of the Google earth engine, and setting a threshold value to exclude areas in which the elevation and the gradient are not consistent.
3. The method for identifying young rubber plantation based on multi-source remote sensing according to claim 1, wherein said determining potential planting area of rubber further comprises:
defining a potential rubber planting area according to the area excluding unsuitable rubber tree planting;
using google near real-time land cover data Dynamic World to exclude areas classified as buildings and bodies of water;
Dynamic World synthetic images of the current year were monitored using the Dynamic World function of geemap packets in python, and the label band of the Dynamic World synthetic images was used to exclude water (label=0) and building (label=6) areas.
4. The method for identifying young rubber plantation based on multi-source remote sensing according to claim 1, wherein said determining potential planting area of rubber further comprises:
Defining a potential rubber planting area according to the land utilization change during the establishment of the plantation;
Monitoring bare soil by using LSWI, and screening a rubber potential planting area with land utilization change according to the condition that the minimum value of the LSWI is less than 0.1;
screening out remote sensing images of 4-12 months in Landsat 8 and Sentinel-2 image sets of corresponding years, and obtaining an LSWI minimum distribution map in the whole target area by using minimum synthesis, namely calling a reduction (ee. Reduction. Min ()) function in a GEE platform, wherein the area with the LSWI minimum value larger than 0.1 is excluded.
5. The method for identifying young rubber plantation based on multi-source remote sensing according to claim 1, wherein said determining potential planting area of rubber further comprises:
potential rubber planting areas are defined according to the historical distribution frequency of rubber planting;
Obtaining a rubber planting distribution frequency chart by superposing a rubber plantation distribution chart of the target area with the step length of 5 years in the past 30 years;
And expanding the area of each rubber planting distribution diagram by using a neighborhood maximum filtering mode, and summing all images to obtain a final rubber planting distribution frequency diagram, so that the area with the planting frequency of 0 is eliminated.
6. The young rubber plantation identification method based on multi-source remote sensing according to claim 1, wherein the synthesizing a time series data set with a certain step length by using a multi-source satellite remote sensing image source according to the potential planting area of the rubber and combining the growth characteristics of rubber seedlings comprises the following steps:
Land resource satellites Landsat 8/9, european empty office sentry No. 2 Sentinel-2 multispectral images and sentry No. 1 Sentinel-1 synthetic aperture radar images are used in GEE; landsat 8/9 and Sentinel-2 TOA optical images have been subjected to radiation and geometry correction using the atmospheric top reflectivity TOA product;
For the Sentinel-2 image, cloud score+ is used to control the Cloud and shadows in the band mask image in the data product;
For Landsat 8/9 images, utilizing clouds and shadows in QA_PIXEL band and QA_ RADSAT band mask images;
Time series of 3 month, 6 month and 12 month steps were synthesized using the median synthesis method, respectively.
7. The young rubber plantation identification method based on multi-source remote sensing according to claim 6, which is characterized in that,
For Landsat 8/9 and Sentinel-2 images, the NDVI and LSWI vegetation indices in each time period were calculated using formulas (1) and (2), respectively:
Wherein ρ Red,ρNIR and ρ SWIR1 are red, near infrared and short wave infrared 1 bands of Landsat 8/9 and Sentinel-2, respectively.
8. The young rubber plantation identification method based on multi-source remote sensing according to claim 7, which is characterized in that,
For Sentinel-2 images, the additional REP index needs to be calculated using equation (3):
Wherein ρ Red1,ρred2 and ρ red3 are red side band 1, red side band 2 and red side band 3 of Sentinel-2 respectively, and are corresponding to B5, B6 and B7 bands of Sentinel-2;
the original spectrum band and the vegetation index band of the time sequence are used as characteristic variables of classification and are input into a classification model.
9. The young rubber plantation identification method based on multi-source remote sensing according to claim 1, wherein the classification using the random forest model to obtain the young rubber plantation distribution map comprises:
The random forest model can be called by using an ee.classifiier.smileRandomForest function in the GEE platform;
Setting the number of decision trees to 100, the remaining settings following default settings in the GEE platform;
Taking the characteristics of the time sequence as input, wherein the shared independent variables of Landsat 8/9 and Sentinel-2 comprise 5 spectral bands and 2 vegetation indexes;
if the random forest model contains Sentinel-2 data, increasing REP vegetation index as an independent variable;
For a random forest model containing Senitnel-1 SAR data, increasing VH and VV polarized radar backscatter coefficients from the set of independent variables;
the dependent variables of each random forest model are binary classifiers to distinguish young rubber gardens (labeled 1) from non-young rubber gardens (labeled 0).
10. The young rubber plantation identification method based on multi-source remote sensing according to claim 9, wherein the classification using the random forest model to obtain the young rubber plantation distribution diagram further comprises:
and inputting all wave bands of each image in the image set synthesized by 3 month step length as a model, predicting a classification label corresponding to each pixel, wherein all pixels with classification labels of 1 are the final distribution area of young rubber.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410134404.3A CN117911783A (en) | 2024-01-31 | 2024-01-31 | Young rubber plantation identification method based on multi-source remote sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410134404.3A CN117911783A (en) | 2024-01-31 | 2024-01-31 | Young rubber plantation identification method based on multi-source remote sensing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117911783A true CN117911783A (en) | 2024-04-19 |
Family
ID=90688935
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410134404.3A Pending CN117911783A (en) | 2024-01-31 | 2024-01-31 | Young rubber plantation identification method based on multi-source remote sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117911783A (en) |
-
2024
- 2024-01-31 CN CN202410134404.3A patent/CN117911783A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Peña-Barragán et al. | Object-based crop identification using multiple vegetation indices, textural features and crop phenology | |
| CN112183209B (en) | Regional crop classification method and system based on multidimensional feature fusion | |
| CN113033670B (en) | Rice planting area extraction method based on Sentinel-2A/B data | |
| CN110852262A (en) | Agricultural land extraction method based on time sequence top-grade first remote sensing image | |
| CN113009485A (en) | Remote sensing tobacco field identification method based on improved vegetation index | |
| CN116091938B (en) | Multisource remote sensing monitoring method for single-cropping rice planting area | |
| CN111209871B (en) | Rape planting land remote sensing automatic identification method based on optical satellite image | |
| Chen et al. | Predicting individual apple tree yield using UAV multi-source remote sensing data and ensemble learning | |
| CN113221806B (en) | Cloud platform fusion multi-source satellite image and tea tree phenological period based automatic tea garden identification method | |
| US20220392215A1 (en) | System and Method for Mapping Land Cover Types with Landsat, Sentinel-1, and Sentinel-2 Images | |
| CN110889394A (en) | Rice lodging recognition method based on deep learning UNet network | |
| KR102496740B1 (en) | System and method for reservoir water body analysis using synthetic aperture radar data | |
| CN114821349B (en) | Forest biomass estimation method taking harmonic model coefficients and climatic parameters into consideration | |
| CN115641412B (en) | Three-dimensional semantic map generation method based on hyperspectral data | |
| CN114926748A (en) | Soybean remote sensing identification method combining Sentinel-1/2 microwave and optical multispectral images | |
| Zhou et al. | An automated, high-performance approach for detecting and characterizing broccoli based on UAV remote-sensing and transformers: A case study from Haining, China | |
| CN114863369B (en) | Method, device, equipment and medium for monitoring corn lodging by laser radar | |
| CN114708490A (en) | Rice planting extraction and multiple cropping index monitoring method, system, terminal and storage medium | |
| CN113534083B (en) | SAR-based corn stubble mode identification method, device and medium | |
| CN113688909A (en) | Ecological region land classification method and device and electronic equipment | |
| CN115063690A (en) | Vegetation classification method based on NDVI (normalized difference vegetation index) time sequence characteristics | |
| CN117911783A (en) | Young rubber plantation identification method based on multi-source remote sensing | |
| Cui et al. | Optimal spatial resolution of remote-sensing imagery for monitoring cantaloupe greenhouses | |
| CN114612794A (en) | Remote sensing identification method for land covering and planting structure in finely-divided agricultural area | |
| CN109918826B (en) | Rice yield estimation method based on key phenological period multi-source vegetation parameters |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |