Sonstiges: |
- Nachgewiesen in: USPTO Patent Applications
- Sprachen: English
- Document Number: 20200225075
- Publication Date: July 16, 2020
- Appl. No: 16/741778
- Application Filed: January 14, 2020
- Claim: 1. A method, comprising: 1) selecting a plurality of sample plots; measuring tree height (H) of each sample plot using a laser hypsometer and tree diameter at breast height (DBH) using a tape; determining a center of each sample plot using a GPS; deriving above-ground biomass (AGB) of vegetation of each individual tree in each sample plot by using species-specific allometric equations with inputs of the tree diameter at breast height and the tree height; adding up the AGB of all individual trees in each sample plot thereby obtaining a total AGB; 2) pre-processing LiDAR data, optical remote sensing data and microwave remote sensing data covering a research region, to yield crown height model (CHM) data, surface reflectance data and a backscattering coefficient, respectively; 3) extracting LiDAR variables based on the CHM data; based on the surface reflectance data, calculating bands between multispectral images using algorithm of band math, and extracting a plurality of optical characteristic vegetation indexes; and extracting a plurality of microwave characteristic variables based on the backscattering coefficient; 4) establishing a multiple stepwise linear regression model of biomass with the measured AGB in the sample plots as a dependent variable and the extracted LiDAR variables as independent variables, bringing the LiDAR data into the linear regression model, and obtaining a biomass estimation value of a coverage region of the LiDAR data; 5) taking the biomass estimation value of the coverage region of the LiDAR data as a training set and a verification sample set, and selecting samples for modeling and verification by a stratified random sampling method; 6) screening out optimal optical characteristic variables and optimal microwave characteristic variables by a variable screening method; and 7) putting the optimal optical characteristic variables, the optimal microwave characteristic variables and a combination of the optimal optical characteristic variables and the optimal microwave characteristic variables to a plurality of prediction models respectively, thereby constructing an optical model, a microwave model, and an optical and microwave synergistic model of AGB retrieval; modeling and verifying according to the samples selected in 5); and selecting an optimal model for biomass retrieval.
- Claim: 2. The method of claim 1, wherein the LiDAR variables in 3) comprise a minimum value, a maximum value, a mean value, a height quantile, a standard deviation, a variation coefficient, a slope, a peak and crown coverage.
- Claim: 3. The method of claim 1, wherein the plurality of optical characteristic vegetation indexes in 3) comprises a normalized difference vegetation index (NDVI), a simple ratio index (SR), an enhanced vegetation index (EVI), a soil adjusted vegetation index (SAVI), a modified soil adjusted vegetation index (MSAVI), an optimized soil adjusted vegetation index (OSAVI), a moisture stress index (MSI), a normalized difference water index (NDWI) and a chlorophyll index (CIgreen), which are calculated based on the following formulas: [mathematical expression included] wherein, R, G, B, NIR and SWIR1 refer to reflectances of red light, green light, blue light, light of near-infrared band and light of short-wave infrared band, respectively.
- Claim: 4. The method of claim 1, wherein the microwave characteristic variables in 3) comprise VV, HH, VH, HV, VV/HH, HH/HV, VV/HV and RVI, and the RVI is calculated based on the following formula: [mathematical expression included] where VV, HH and VH are polarization modes, and VV/HH, HH/HV and VV/HV are ratios.
- Claim: 5. The method of claim 1, wherein the variable selection method adopted in 6) is a stepwise screening method.
- Claim: 6. The method of claim 5, wherein the optimal optical characteristic variables in 6) comprise SR, NDVI, OSAVI, MSI and NDWI, and the optimal microwave characteristic variables comprise VV, HV, HH/HV and RVI.
- Claim: 7. The method of claim 1, wherein the plurality of prediction models in 7) comprises models of multiple stepwise linear regression (SLR), K nearest neighbor (KNN), a support vector machine (SVM), a BP neural network (BPNN), a random forest (RF) and deep learning (DL), wherein a Stacked Sparse Autoencoder network (SSAE) model is selected as a deep learning model.
- Claim: 8. A system, comprising: a first module, configured to select a plurality of sample plots; measure tree height (H) of each sample plot using a laser hypsometer and tree diameter at breast height (DBH) using a tape; determine a center of each sample plot using a GPS; derive above-ground biomass (AGB) of vegetation of each individual tree in each sample plot by using species-specific allometric equations with inputs of the tree diameter at breast height and the tree height; add up the AGB of all individual trees in each sample plot thereby obtaining a total AGB; a second module, configured to pre-process LiDAR data, optical remote sensing data and microwave remote sensing data covering a research region, to yield crown height model (CHM) data, surface reflectance data and a backscattering coefficient, respectively; a third module, configured to extract LiDAR variables based on the CHM data; based on the surface reflectance data, calculate bands between multispectral images using algorithm of band math, and extract a plurality of optical characteristic vegetation indexes; and extract a plurality of microwave characteristic variables based on the backscattering coefficient; a fourth module, configured to establish a multiple stepwise linear regression model of biomass with the measured AGB in the sample plots as a dependent variable and the extracted LiDAR variables as independent variables, bring the LiDAR data into the linear regression model, and obtain a biomass estimation value of a coverage region of the LiDAR data; a fifth module, configured to take the biomass estimation value of the coverage region of the LiDAR data as a training set and a verification sample set, and select samples for modeling and verification by a stratified random sampling method; a sixth module, configured to screen out optimal optical characteristic variables and optimal microwave characteristic variables by a variable screening method; and a seventh module, configured to put the optimal optical characteristic variables, the optimal microwave characteristic variables and a combination of the optimal optical characteristic variables and the optimal microwave characteristic variables to a plurality of prediction models respectively, thereby constructing an optical model, a microwave model, and an optical and microwave synergistic model of AGB retrieval; model and verify according to the samples; and select an optimal model for biomass retrieval.
- Current International Class: 01; 06; 06; 01; 06
|