CN-121999388-A - Arbor-shrub mixed pixel vegetation index time sequence data decomposition method

CN121999388ACN 121999388 ACN121999388 ACN 121999388ACN-121999388-A

Abstract

The invention discloses a arbor and shrub mixed pixel vegetation index time sequence data decomposition method which comprises the steps of obtaining target vegetation community coarse resolution mixed pixel vegetation index time sequence data, high resolution vegetation index time sequence data and arbor and shrub distribution data, forming decomposed time constraint conditions based on fluctuation ranges of the high resolution data statistics arbor and shrub greenness and difference between arbor and shrub greenness, embedding the time constraint conditions into a linear mixed model, solving vegetation index reference time sequence of arbor and shrub end members in a target area, and weighting and summing residual errors of mixed pixel vegetation indexes and arbor and shrub reference values according to area occupation ratio, wherein the arbor and shrub are oriented to pixel and time sequence adjustment weights. The method can decompose the vegetation index time sequence data of the mixed pixels into arbor and shrub vegetation index time sequence data of sub-pixel levels, the decomposition result is reliable, the community mobility is good, and effective technical support can be provided for the fine monitoring of the sub-pixel level vegetation.

Inventors

ZHU WENQUAN
MENG GE
ZHANG HUI

Assignees

北京师范大学

Dates

Publication Date: 20260508
Application Date: 20260206

Claims (9)

1. A arbor-shrub mixed pixel vegetation index time sequence data decomposition method is characterized by comprising the following steps: Acquiring target vegetation community coarse resolution mixed pixel vegetation index time sequence data, high resolution vegetation index time sequence data with spatial resolution superior to the coarse resolution data, and distribution data of arbor and shrub and soil background; Based on high-resolution data, calculating greenness of arbor-shrub and soil background and fluctuation range of difference between arbor-shrub and soil background, and forming a decomposed time constraint condition; embedding time constraint conditions into the linear mixed model, and calculating a vegetation index reference time sequence of the arbor-shrub grass and soil background end member in the target area; And (3) distributing the mixed pixel vegetation index and the arbor-shrub and soil background reference value to arbor-shrub vegetation index time sequence curves of sub-pixel level by the residual error of weighted summation according to the area occupation ratio and the arbor-shrub and soil background facing pixel and time sequence adjustment weights.
2. The arbor-shrub mixed pixel vegetation index time sequence data decomposition method of claim 1, wherein the time constraint condition is constructed by calculating the average value of arbor-shrub, grass and soil background components in a region as component end member vegetation index values in corresponding periods for high-resolution vegetation index time sequence data in a period-by-period mode, dividing a year period into 36 sections according to ten days, counting the maximum value and the minimum value of arbor-shrub, grass and soil background vegetation indexes in each period as upper limit and lower limit constraints of a calculation result in corresponding periods, and simultaneously counting the maximum value and the minimum value of arbor-shrub, shrub and soil background two component vegetation index difference values, and constraining the relative sizes of the component end member vegetation index calculation results in corresponding periods.
3. The arbor and shrub mixed pixel vegetation index time series data decomposition method according to claim 1, wherein the arbor and shrub vegetation index reference time series is calculated based on a linear mixed model principle, and a calculation formula is: Wherein, the Is the vegetation index of the coarse resolution mixed pixel, Is the end member vegetation index of component c, Representing the area ratio of component c in the picture element.
4. The method of claim 3, wherein the vegetation index of the plurality of mixed pixels is known in the target area, the number of pixels is greater than the number of components, and the vegetation index of each end member of the arbor, the shrub, the grass and the soil is calculated by combining the area ratio of each component of the arbor, the shrub, the grass and the soil, and the vegetation index of each end member calculated by the period forms a reference time sequence of the vegetation index in the target area.
5. The arbor and shrub mixed pixel vegetation index time sequence data decomposition method of claim 3, wherein the end member vegetation index solution is implemented by a genetic algorithm, and for a mixed pixel sample set in a target area, an optimal solution combination of each component end member vegetation index is determined through repeated iterative searching, so that root mean square error between a predicted value and an observed value of all mixed pixel sample vegetation indexes is minimum, and the formula is as follows: Where n is the number of samples of the mixed pixel, For the indexing of the picture elements, Is the first Observations of the vegetation indices of the individual hybrid pixels, Is calculated according to a linear mixed model formula according to candidate end member vegetation index and area ratio Vegetation index predictive value of each pixel, when And when the minimum value is taken, the component end member vegetation index is obtained.
6. The method for decomposing time series data of arbor-shrub mixed pel vegetation index according to claim 5, wherein when the genetic algorithm solves arbor-shrub, grass and component end member vegetation indexes each time period by time period, judging the attribution of solving dates in 36 time periods divided by ten days in the annual period, and restraining the solving results according to time constraint conditions of the corresponding time periods.
7. The method of claim 6, wherein the step-by-step calculation of a weighted sum of the arbor, shrub, grass, soil component reference vegetation indices and the respective area ratios, and calculating a residual of the mixed pel vegetation indices minus the weighted sum.
8. The arbor and shrub mixed pixel vegetation index time sequence data decomposition method of claim 1, wherein the residual allocation is based on two factors of component area occupation ratio and representative of a target area vegetation index reference value to the component vegetation index in the mixed pixel, and a composite weight is constructed and normalized, and a calculation formula is as follows: Wherein, the Is one of the components of the arbor and shrub and the soil, As the weight of the component(s), Is the area ratio of the components, The root mean square error of the components is obtained by multiplying the area ratio of the four components by the root mean square error, and the denominator is used for normalizing the weight.
9. The method of claim 8, wherein in the component weight calculation, statistics are performed on the arbor, shrub, grass and soil components based on the high-resolution vegetation index, and root mean square error of the high-resolution vegetation index and the regional end member vegetation index of a single component in the range of the coarse-resolution mixed pixels is calculated to measure the difference between the vegetation index of the component in the pixels and the regional vegetation index reference value.

Description

Arbor-shrub mixed pixel vegetation index time sequence data decomposition method Technical Field The invention belongs to the field of vegetation remote sensing, and particularly relates to a arbor and shrub mixed pixel vegetation index time sequence data decomposition method. Background The vegetation index time sequence data can represent the dynamic characteristics of vegetation green changes along with time, reveal vegetation growth and evolution rules, and are key data sources in ecological environment monitoring. The vegetation index time sequence data is widely applied to remote sensing vegetation climate monitoring, can provide basis for community ecological niches and ecological system function research, and is also a key parameter for biomass estimation, carbon reserve estimation and land ecological system productivity simulation. Under the climate change background, the vegetation index time sequence data is also beneficial to analyzing the response mechanism of vegetation to climate factors such as temperature, precipitation and the like, determining the effect of vegetation in carbon circulation and water circulation, and providing important basis for verification and calibration of climate models. However, the vegetation index time sequence data obtained by remote sensing is essentially a mixed signal of arbor and shrub vegetation and soil background in the pixels, and the pixels are regarded as a whole for analysis in the prior art, so that dynamic characteristics of arbor and shrub on the sub-pixel scale are not described. The method leads to the difficulty in distinguishing the difference of the weathers of each vegetation component based on the pixel-level data, so that the weathers parameters of the remote sensing inversion are difficult in ecological interpretation, the attribution of the vegetation type of the total signal cannot be clarified, and the requirement of fine ecological monitoring is difficult to meet. Disclosure of Invention Aiming at the technical defect that the existing mixed pixel vegetation index time sequence data is difficult to finely describe the independent dynamic characteristics of various vegetation components of arbor, shrub and grass in the sub-pixel scale, the invention provides the arbor, shrub and grass mixed pixel vegetation index time sequence data decomposition method, on the basis of space constraint of a traditional linear mixed model, a time constraint condition is added, and the mixed pixel vegetation index time sequence data is accurately decomposed into the sub-pixel arbor, shrub and grass vegetation index time sequence data through the multidimensional constraint collaborative optimization decomposition process, so that a reliable decomposition result with good community mobility is obtained. In order to achieve the above purpose, the invention provides a arbor and shrub mixed pixel vegetation index time sequence data decomposition method, which comprises the following steps: acquiring coarse-resolution mixed pixel vegetation index time sequence data of a target vegetation community, high-resolution vegetation index time sequence data and distribution data of arbor, shrub and soil background; Based on high-resolution data, calculating greenness of arbor-shrub and soil background and fluctuation range of difference between arbor-shrub and soil background, and forming a decomposed time constraint condition; Embedding time constraint conditions in the linear mixed model, and calculating a vegetation index reference time sequence of the arbor-shrub and soil background end members (namely end members, categories comprise soil backgrounds to adapt to pixel all-land feature components, and the core calculation target is arbor-shrub); And (3) distributing the mixed pixel vegetation index and the arbor-shrub and soil background reference value to arbor-shrub vegetation index time sequence curves of sub-pixel level by the residual error of weighted summation according to the area occupation ratio and the arbor-shrub and soil background facing pixel and time sequence adjustment weights. The time constraint condition is constructed by calculating the average value of arbor, shrub, grass and soil background components (based on arbor and shrub distribution data of high resolution) in a region as component end member vegetation index values of corresponding periods for time sequence data of the high resolution vegetation index, dividing a year period into 36 sections according to ten days, counting the maximum value and the minimum value of the arbor, shrub, grass and soil background vegetation index of the high resolution in each period as upper and lower limit constraints of a solution result of the corresponding period, and counting the maximum value and the minimum value of arbor, shrub, grass and soil background two component vegetation index difference values at the same time, and restraining the relative sizes of the end member vegetation index solution