CN-121999365-A - Grassland biomass and carbon fixation remote sensing estimation method and system thereof

CN121999365ACN 121999365 ACN121999365 ACN 121999365ACN-121999365-A

Abstract

The application relates to a grassland biomass and carbon sequestration remote sensing estimation method and system. The method comprises the steps of obtaining a hyperspectral remote sensing image, carrying out atmospheric correction to obtain a ground surface reflectivity image, identifying a grassland functional end member to obtain a localized grassland end member spectrum library, carrying out linear spectrum mixing decomposition to obtain a grassland plant functional abundance set by combining the ground surface reflectivity image and the localized grassland end member spectrum library, calculating pixel-level dynamic specific carbon content by combining time sequence multispectral image data to obtain a dynamic specific carbon content field, carrying out regression prediction calculation to obtain a grassland biomass distribution map by combining the ground surface reflectivity image, radar data and the grassland plant functional abundance set, and carrying out pixel-by-pixel multiplication to obtain overground fixed carbon volume spatial distribution data. The method can solve the problem of estimation deviation caused by using the fixed carbon content coefficient in the prior art, and improves the estimation precision.

Inventors

ZHANG FUCUN
LIU JIARU
LI HONGYING
HE JUHONG
XU SHANSHAN
Yan Luqing

Assignees

青海理工学院

Dates

Publication Date: 20260508
Application Date: 20260122

Claims (10)

1. A remote sensing estimation method for grassland biomass and carbon sequestration thereof, the method comprising: s1, acquiring a hyperspectral remote sensing image covering a target grass area, and performing atmospheric correction calculation according to the hyperspectral remote sensing image to obtain an earth surface reflectivity image; S2, carrying out grassland functional end member identification according to the surface reflectivity image to obtain a localized grassland end member spectrum library; s3, performing linear spectrum mixed decomposition according to the surface reflectivity image and the localized grassland end member spectrum library to obtain a grassland plant functional abundance set; s4, calculating the pixel-level dynamic specific carbon content according to the grassland plant functional abundance set and combining the time sequence multispectral image data to obtain a dynamic specific carbon content field; s5, carrying out regression prediction calculation according to the surface reflectivity image, radar data and the grassland plant functional abundance set to obtain a grassland biomass distribution map; And S6, performing pixel-by-pixel multiplication calculation according to the grassland biomass distribution map and the dynamic specific carbon content field to obtain the spatial distribution data of the ground carbon fixation amount.
2. The method according to claim 1, wherein S1 comprises: S11, acquiring radiation calibration parameters according to the hyperspectral remote sensing image, and performing radiation calibration calculation on the radiation calibration parameters to obtain apparent radiance; S12, acquiring the atmospheric parameters of the target grassland area, and performing atmospheric correction calculation through a 6S radiation transmission model according to the apparent radiance and the atmospheric parameters to obtain the surface reflectivity image.
3. The method according to claim 1, wherein S2 comprises: s21, performing noise separation transformation on the surface reflectivity image to obtain a transformed main component; S22, calculating a pure pixel index according to the principal component, and extracting pixels with the pure pixel index larger than a preset threshold value to obtain candidate end member pixels; S23, projecting the candidate end member pixels to obtain projection data; S24, rotating the projection data, and identifying pixels located at the top of the convex surface to obtain pure end member pixels; S25, calibrating the grassland functional end members according to the spectral characteristics of the pure end member pixels to obtain calibrated grassland functional end members, wherein the grassland functional end members comprise grass end members, nutgrass end members, leguminous end members, dead end members and bare earth end members; S26, carrying out average value calculation on end member spectrums of the same category in the calibrated grassland functional end members to obtain average spectrums corresponding to the functional end members; And S27, combining the average spectrums corresponding to the functional end members to obtain the localized grassland end member spectrum library.
4. A method according to claim 3, wherein said S3 comprises: s31, constructing a linear spectrum mixing equation set with constraint according to the spectrum vector of the grassland pixels in the surface reflectivity image and the spectrum vector of each end member in the localized grassland end member spectrum library; S32, carrying out full constraint least square method solution on the linear spectrum mixing equation set with the constraint to obtain the abundance of grass end members, the abundance of nutgrass flatsedge end members and the abundance of leguminous end members corresponding to each grassland pixel; S33, arranging and combining the abundance of the grass end members, the abundance of the cyperus end members and the abundance of the leguminous end members according to pixel positions to obtain the grassland plant functional abundance set.
5. The method of claim 4, wherein S4 comprises: s41, constructing a normalized vegetation index time sequence curve according to the time sequence multispectral image data to obtain a vegetation index time sequence curve; S42, respectively extracting the annual maximum normalized vegetation index value and the current time normalized vegetation index value corresponding to the grass, the nutgrass flatsedge and the leguminous plant based on the vegetation index time sequence curve to obtain the annual maximum normalized vegetation index value and the current time normalized vegetation index value corresponding to the grass, the nutgrass flatsedge and the leguminous plant; S43, calculating according to the maximum normalized vegetation index value in the year and the normalized vegetation index value at the current moment to obtain the weather correction factors of the corresponding grasses, nutgrass flatsedge and leguminosae, wherein the weather correction factors have the calculation formula: Wherein, the Is the first The plant function is at the moment Is used for the object-weather modification factor of (c), Is the first The plant function is at the moment Is used for normalizing the vegetation index value at the current moment, Is the first The plant function is at the moment The maximum normalized vegetation index value within the year of the year, Is the first Plant functional type carbon accumulation regulation coefficient in the dry and yellow period; s44, calculating the dry matter content density of the pixel-level blade according to the surface reflectivity image to obtain a dry matter content density map of the pixel-level blade; S45, carrying out weighted summation calculation on the searched preset functional basis specific carbon content coefficients according to the grassland plant functional abundance set, the climatic correction factors and the pixel-level leaf dry matter content density map to obtain a pixel-level dynamic specific carbon content value, wherein the weighted summation calculation formula is as follows: Wherein, the Is a picture element At the moment of time A pel level dynamic specific carbon content value of (c), Is a picture element Middle (f) The abundance of the functional end members of the plant, Is the first Said base specific carbon content coefficient of the plant function, Is a picture element At the moment of time Is used for the dry matter content density of the blade, Is the first The plant function is at the moment Is a factor of the climate correction; s46, arranging the pixel-level dynamic specific carbon content values according to pixel positions to obtain the dynamic specific carbon content field.
6. The method of claim 4, wherein S5 comprises: s51, calculating vegetation index features according to the surface reflectivity images to obtain a vegetation index feature set; S52, resampling the radar data to obtain resampled radar data, wherein the radar data are data matched with the surface reflectivity image; S53, extracting a backscattering coefficient of the resampled radar data to obtain a radar feature set; S54, splicing the vegetation index feature set, the radar feature set and the grassland plant functional abundance set to obtain a comprehensive feature vector set; S55, predicting the comprehensive feature vector set through a random forest regression model to obtain the grassland biomass distribution map.
7. The method according to claim 1, wherein S6 comprises: S61, extracting a biomass estimated value of each pixel in the biomass distribution map on the grassland and a pixel-level dynamic specific carbon content value of a corresponding pixel in the dynamic specific carbon content field; S62, multiplying the biomass estimated value and the pixel-level dynamic specific carbon content value to obtain a pixel-level ground carbon fixation value; And S63, arranging the pixel-level ground carbon sequestration amount values according to pixel positions to obtain the ground carbon sequestration amount spatial distribution data.
8. A remote sensing estimation system for grassland biomass and carbon sequestration thereof, the system comprising: the atmosphere correction module is used for acquiring a hyperspectral remote sensing image covering a target grassland area, and performing atmosphere correction calculation according to the hyperspectral remote sensing image to obtain an earth surface reflectivity image; The grassland end member identification module is used for identifying grassland functional end members according to the surface reflectivity image to obtain a localized grassland end member spectrum library; the spectrum mixing and decomposing module is used for carrying out linear spectrum mixing and decomposing according to the surface reflectivity image and the localized grassland end member spectrum library to obtain a grassland plant functional abundance set; The dynamic specific carbon content calculation module is used for calculating pixel-level dynamic specific carbon content according to the grassland plant functional abundance set and by combining time sequence multispectral image data to obtain a dynamic specific carbon content field; the aboveground biomass prediction module is used for carrying out regression prediction calculation according to the surface reflectivity image, the radar data and the grassland plant functional abundance set to obtain a grassland aboveground biomass distribution map; And the carbon sequestration amount calculation module is used for carrying out pixel-by-pixel multiplication calculation according to the grassland biomass distribution map and the dynamic specific carbon content field to obtain the space distribution data of the overground carbon sequestration amount.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the method of any one of claims 1 to 7 when executing the computer program.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of any of claims 1 to 7.

Description

Grassland biomass and carbon fixation remote sensing estimation method and system thereof Technical Field The invention belongs to the technical field of ecology, and particularly relates to a grassland biomass and carbon fixation remote sensing estimation method and system. Background With the rapid development of remote sensing technology, estimation of regional grassland biomass and carbon sequestration by using the remote sensing technology has become an important means for grassland ecological research and global change research. The grassland ecological system is used as an important component of the land ecological system, plays a key role in regulating global carbon balance and maintaining ecological balance, and has important significance in accurately evaluating the carbon fixing capability of the grassland ecological system for coping with climate change and making reasonable ecological management policy. In the conventional technology, an optical remote sensing image is generally adopted to calculate a vegetation index, and a statistical regression model is established in combination with ground measured data to estimate the above-ground biomass. After biomass is obtained, a fixed carbon content coefficient (conversion coefficient) is obtained by consulting a literature or field measurement, and the coefficient is multiplied with the estimated biomass, so that the regional grassland carbon fixation is calculated. The method mainly relies on optical remote sensing data of a single time phase, and the ecological parameters of the regional scale are extrapolated through an empirical model. However, the traditional method generally uses a fixed specific carbon content coefficient for calculation, ignores the dynamic change characteristics of carbon accumulation of different plant functional types in different growth stages and climatic periods, and results in that the estimation result of the carbon fixation amount cannot reflect the actual carbon conversion process and has lower precision. Meanwhile, the existing estimation means are difficult to effectively solve the problem of mixed pixels under the complex ground surface background, and the contribution of different grassland components cannot be finely distinguished from the pixel scale, so that the estimation deviation of the spatial distribution of the ground biomass and the carbon fixation is larger in the region with stronger grassland vegetation heterogeneity, and the requirement of accurate ecological monitoring cannot be met. Disclosure of Invention Based on the above, it is necessary to provide a remote sensing estimation method and system for the grassland biomass and the carbon fixation thereof, which can realize pixel-level dynamic inversion and improve estimation accuracy. In a first aspect, the application provides a remote sensing estimation method for grassland biomass and carbon sequestration thereof, comprising: S1, acquiring a hyperspectral remote sensing image covering a target grass area, and performing atmospheric correction calculation according to the hyperspectral remote sensing image to obtain an earth surface reflectivity image; s2, carrying out grassland functional end member identification according to the surface reflectivity image to obtain a localized grassland end member spectrum library; S3, carrying out linear spectrum mixed decomposition according to the surface reflectivity image and a localized grassland end member spectrum library to obtain a grassland plant functional abundance set; s4, calculating the pixel-level dynamic specific carbon content according to the grassland plant functional abundance set and combining the time sequence multispectral image data to obtain a dynamic specific carbon content field; S5, carrying out regression prediction calculation according to the surface reflectivity image, the radar data and the grassland plant functional abundance set to obtain a grassland biomass distribution map; and S6, performing pixel-by-pixel multiplication calculation according to the grassland biomass distribution map and the dynamic specific carbon content field to obtain the spatial distribution data of the overground carbon fixation amount. In one embodiment, S1 comprises: S11, acquiring radiation calibration parameters according to the hyperspectral remote sensing image, and performing radiation calibration calculation on the radiation calibration parameters to obtain apparent radiance; S12, acquiring the atmospheric parameters of the target grassland area, and performing atmospheric correction calculation through a 6S radiation transmission model according to the apparent radiance and the atmospheric parameters to obtain an earth surface reflectivity image. In one embodiment, S2 comprises: S21, performing noise separation transformation on the surface reflectivity image to obtain a transformed main component; S22, calculating a pure pixel index according to the main component components, and extr