CN113340898A - Leaf area index space-time change characteristic research method - Google Patents

Leaf area index space-time change characteristic research method Download PDF

Info

Publication number
CN113340898A
CN113340898A CN202110731055.XA CN202110731055A CN113340898A CN 113340898 A CN113340898 A CN 113340898A CN 202110731055 A CN202110731055 A CN 202110731055A CN 113340898 A CN113340898 A CN 113340898A
Authority
CN
China
Prior art keywords
lai
area
monitored
pixel
vegetation
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.)
Granted
Application number
CN202110731055.XA
Other languages
Chinese (zh)
Other versions
CN113340898B (en
Inventor
沙晋明
高尚
李小梅
吴龙滨
沙昱
王金亮
包忠聪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fujian Normal University
Original Assignee
Fujian Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Fujian Normal University filed Critical Fujian Normal University
Priority to CN202110731055.XA priority Critical patent/CN113340898B/en
Publication of CN113340898A publication Critical patent/CN113340898A/en
Application granted granted Critical
Publication of CN113340898B publication Critical patent/CN113340898B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N2021/1765—Method using an image detector and processing of image signal
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N2021/178—Methods for obtaining spatial resolution of the property being measured
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
    • G01N2021/1793—Remote sensing
    • G01N2021/1797—Remote sensing in landscape, e.g. crops
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N2021/8466—Investigation of vegetal material, e.g. leaves, plants, fruits
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30181—Earth observation
    • G06T2207/30188—Vegetation; Agriculture
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

The invention discloses a leaf area index space-time change characteristic research method, which comprises the following implementation steps: 1) acquiring a remote sensing image and field actual measurement data of an area to be monitored, and constructing an LAI inversion model to obtain LAI time sequence data of the area to be monitored; 2) analyzing the space-time change of the area to be monitored, analyzing the space fluctuation of the LAI of the area to be monitored by using a variation coefficient pixel by pixel, analyzing the space change trend of the LAI by pixel by a unary linear regression analysis method, analyzing the sustainability of the LAI change by pixel by an R/S analysis method, and analyzing the relationship between the LAI and the LAI change trend type and the terrain factor by combining terrain data; 3) carrying out land cover classification on remote sensing image data of an area to be monitored by using a supervision classification method, introducing a GWR (global warming potential) model to analyze the relation between different land cover types and LAI (local area identity) so as to obtain the relation between different land cover type patterns and the LAI of the area to be monitored; the method has the characteristics of reasonable model construction, high inversion accuracy, reasonable estimation result and wide application range.

Description

Leaf area index space-time change characteristic research method
Technical Field
The invention relates to the field of leaf area index research, in particular to a method for researching space-time change characteristics of a leaf area index.
Background
The Leaf Area Index (LAI) is one of important structural parameters of vegetation, can represent the density degree and the canopy structural characteristics of leaves, is closely related to the biophysical processes of photosynthesis, respiration, transpiration and the like of the vegetation and the carbon, nitrogen, water circulation and the like of the earth ecosystem, and the change of the LAI is an important Index for estimating the vegetation coverage and monitoring and forecasting the growth vigor, biomass, yield and the like of crops, so that the rapid and accurate estimation of the LAI has important research significance for agricultural monitoring, biological geochemical cycle and the like.
At present, the LAI inversion method based on the remote sensing technology mainly comprises an empirical model method based on a vegetation index and a physical model method based on a radiation transmission model. The two models are essentially distinct but complementary. The physical model is strong in mechanism, the used vegetation type and the space range are wider, but the model parameters are more and difficult to obtain, and the problem of ill-condition of model inversion exists. The empirical model based on the vegetation index is an important means for coupling remote sensing observation and ground observation data, can monitor the growth vigor of crops timely, effectively and harmlessly, but is weak in mechanism and easy to be influenced by vegetation types, regions and the like.
Disclosure of Invention
In view of the defects in the background art, the technical problem to be solved by the invention is to provide a leaf area index space-time change characteristic research method which has the characteristics of reasonable model construction, high inversion precision, reasonable estimation result, good universality and wide application range.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a method for researching the space-time change characteristics of a leaf area index comprises the following implementation steps:
1) acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, constructing an LAI inversion model, inverting LAI time series data of the region to be monitored, obtaining the LAI time series data of the region to be monitored, and carrying out precision verification;
2) performing space-time change analysis on the area to be monitored, after LAI time sequence data of the area to be monitored is obtained, analyzing the space volatility of the LAI of the area to be monitored by using a variation coefficient pixel by pixel, analyzing the space change trend of the LAI pixel by a unary linear regression analysis method, analyzing the sustainability of the change of the LAI pixel by an R/S analysis method, and analyzing the relation between the type of the change trend of the LAI and a terrain factor by combining terrain data;
3) carrying out land cover classification on remote sensing image data of an area to be monitored by using a supervision classification method, counting average pixel LAI values of various land cover types and benefits of land cover change on total LAI, introducing a GWR (global warming potential) model to analyze the relation between different land cover types and the LAI and the spatial non-stationarity of the different land cover types and the LAI so as to reveal the spatial non-stationarity relation between the land cover and the area to be monitored, and obtaining the relation between different land cover type patterns and the area to be monitored LAI;
as a further improvement of the technical scheme of the invention:
the method comprises the following steps of 1), acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, and constructing an LAI inversion model, wherein the detailed step of inverting LAI time sequence data of the region to be monitored comprises the following steps:
A1) when field actual measurement data of an area to be monitored is obtained, sampling and selecting a 30m multiplied by 30m sample plot for investigation, selecting a place with uniform vegetation distribution and representativeness as the sample plot, setting sampling points at four corners and the center of each sample plot, carrying out vegetation canopy leaf area index collection on each sampling point by using an LAI-2000 plant canopy analyzer, and finally taking the average value of the sampling points in a sample square as the LAI value of the sample square to obtain the LAI value of the sample square;
A2) when an inversion model is constructed, 2 geographic environment factors of vegetation coverage and elevation values are introduced as key factors of the inversion model, the key factors and sensitive bands of LAI are used as model input parameters together, a multiple linear regression model of the LAI sensitive bands, the geographic environment factors and actually measured LAI is constructed by using remote sensing image data and field sample data, and LAI time sequence data of an area to be monitored is obtained;
introducing vegetation coverage and 2 geographic environment factors with elevation values as key factors of an inversion model in the step A2), specifically, taking the vegetation coverage as an input parameter to participate in construction of the inversion model, inverting the vegetation coverage by using a pixel binary model based on NDVI (normalized difference of variance), and calculating the vegetation coverage according to the difference of the vegetation coverage and the elevation value
Figure DEST_PATH_IMAGE001
Obtaining vegetation coverage, wherein NDVIsoil is the NDVI value of a region which is completely bare or not covered by vegetation, and NDVIveg is the NDVI value of a region which is completely covered by plants, and a method for determining NDVIsoil and NDVIveg is provided by the first plum seedling of the author;
in the step 2), the spatial volatility of the LAI of the area to be monitored is analyzed pixel by using the variation coefficient, the spatial variation trend of the LAI is analyzed pixel by a unary linear regression analysis method, and the detailed step of analyzing the sustainability of the LAI variation pixel by an R/S analysis method comprises the following steps:
B1) performing pixel-by-pixel stability analysis on LAI data of the region to be monitored, and evaluating the stability of LAI along with time by using the variation coefficient according to
Figure 873999DEST_PATH_IMAGE002
Obtaining the fluctuation condition of vegetation LAI of an area to be monitored, wherein n represents the number of years of monitoring, xi represents the annual LAI value, x represents the average value of the LAI, the larger CV is, the more scattered the data is, the smaller CV is, the more compact the data is, and the more stable the vegetation change is;
B2) quantitatively researching the change trend of vegetation LAI of the area to be monitored by adopting a unary linear regression analysis method pixel by pixel, wherein the obtained value is the slope of the unary linear regression analysis of each pixel in the monitoring year
Figure DEST_PATH_IMAGE003
Obtaining the variation trend of the region to be monitored, wherein n is the total year of variation monitoring, Li is the value of LAI in the ith year, and SLOPE is the SLOPE of unary linear regression analysis;
B3) the sustainability of LAI change is analyzed pixel by adopting an R/S analysis method, the H value calculated by the R/S analysis method, namely the Hurst index can better reflect the autocorrelation of time sequence data, and the sustainability of the LAI change trend of vegetation in the area to be monitored pixel by pixel can be obtained by combining the Hurst index obtained by the analysis result of the R/S analysis method with the result of the change trend analysis;
the detailed step that the sustainability of the pixel-by-pixel variation trend of vegetation LAI of the area to be monitored can be obtained by combining the Hurst index obtained from the analysis result of the R/S analysis method in the step B3) with the variation trend analysis result comprises the following steps:
B31) according to
Figure 54314DEST_PATH_IMAGE004
Calculating a differential sequence of the LAI of the area to be monitored according to the differential sequence
Figure DEST_PATH_IMAGE005
Obtaining a mean sequence of the LAI;
B32) according to
Figure 746326DEST_PATH_IMAGE006
Obtaining the dispersion of the area to be monitored according to
Figure DEST_PATH_IMAGE007
Obtaining the range of the region to be monitored, and finally obtaining the range according to
Figure 670289DEST_PATH_IMAGE008
Obtaining the standard deviation, the Hurst index is the ratio of R (m)/S (m), when 0<H<When the time sequence data of the LAI is 0.5, the trend change characteristics before and after the time sequence data of the LAI are opposite, namely the change of the LAI has reverse persistence, and the closer the H index is to 0, the stronger the reverse persistence is; when 0.5<H<When 1, the LAI time sequence data has consistent trend before and after, namely the change of the LAI has sustainabilityContinuity, and the closer the H index is to 1, the stronger the sustainability is; when H =0.5, the change of the LAI time series data before and after is random, that is, the change trend cannot be judged;
analyzing the relationship between the LAI and the LAI variation trend type and the terrain factor in the step 2) by combining the terrain data, specifically, eliminating the terrain area difference by using a terrain area correction coefficient K to reveal the distribution and the evolution trend of the LAI variation type on different terrain features, according to the distribution and the evolution trend
Figure DEST_PATH_IMAGE009
Obtaining a K value, wherein delta Vi is the area of a certain LAI change type under a certain topographic feature, and delta V is the area of a certain vegetation type in the whole research area; ti is the area of a particular topographical feature; s is the area of the whole study area;
the detailed step of introducing a GWR model in the step 3) to analyze the relationship between different land cover types and the LAI and the spatial non-stationarity of the land cover types and obtain the relationship between different land cover type patterns and the LAI of the area to be monitored comprises the following steps:
C1) the GWR model is based on the first law of geography, adds the geographic location of the data to the regression parameters of the model on the basis of an OLS (orientation left Square) model, taking into account the spatial weights of neighboring points, allowing local rather than global parameter estimation, based on which
Figure 865778DEST_PATH_IMAGE010
Obtaining a dependent variable of a sample point i, wherein yiIs a dependent variable, beta, of a sample point i0Is a constant term of regression, betak As regression parameter, xikIs the k-th variable observed for sample point i, εiFor error term, (mu)i，νi) Is the geographic coordinate, β, of the sample point i0（μi，νi) As the regression constant term, beta, for the sample point ik（μi，νi) Is xkRegression parameters at sample points i;
C2) according to
Figure DEST_PATH_IMAGE011
Calculating betak（μi，νi) Wherein X is a matrix of independent variables, XTIs the transpose of matrix X, Y is the dependent variable matrix, W (mu)i，νi) A matrix that is a spatial weight;
C3) according to
Figure 482573DEST_PATH_IMAGE012
Calculating a matrix of spatial weights, wherein WijTo estimate the weight of the point i to be measured using the spatial sample point j, dijAnd h is the bandwidth, and the Euclidean distance between the sampling point j and the point i to be estimated is determined by utilizing the minimum Chi information criterion.
The invention has the following advantages:
1. according to the invention, 2 key geographic environment factors of the elevation value and the vegetation coverage are introduced into the inversion model to serve as inversion constraint conditions, so that the accuracy of the Leaf Area Index (LAI) inversion model is obviously improved, and the relationship between different land coverage types and the LAI is revealed by introducing the geographic weighted regression model.
2. The method comprises the steps of measuring ground actual measurement data through an LAI-2000 vegetation canopy analyzer, selecting a sample plot of 30m multiplied by 30m for investigation during sampling, selecting a place with uniform vegetation distribution and representativeness as the sample plot, and finally taking the average value of sampling points in the sample plot as the LAI value of the sample plot to verify the accuracy of an inversion model, wherein the inversion accuracy is accurate and reliable.
3. In the aspect of sensitivity analysis, a global sensitivity analysis algorithm Sobol based on variance is selected for calculating first-order sensitivity and total sensitivity, compared with the original algorithm, the improved Sobol algorithm has stronger stability and lower calculation cost, and the defect that mutual influence among parameters is neglected in local analysis of parameter sensitivity is overcome.
4. The invention adopts an R/S analysis method to quantitatively describe the dependency of time sequence information, the calculated Hurst index can better reflect the autocorrelation of time sequence data, and the sustainability of the variation trend of vegetation LAI from pixel to pixel in a research area can be obtained by combining the result of the variation trend analysis, so that the invention is completely feasible in technology and can better judge the variation trend.
5. The invention eliminates the irrationality of distribution characteristics of LAI variation types under specific terrain conditions by using the terrain area correction coefficient K, and avoids the unreasonable influence on analysis results caused by the fact that certain LAI variation types occupy a large area on certain terrain characteristics but occupy a small area in the whole research area.
6. According to the method, the geographical weighted regression model GWR capable of reflecting the relation characteristics between local geographical elements is used for revealing the spatial non-stationarity relation between the LAI and different land cover types, so that the defect that a unitary linear regression model reflects the local relation is overcome, and different land cover types on different spatial positions have different influence modes and degrees on the LAI.
7. According to the invention, by remote sensing and GIS technology, aiming at the problem that when LAI inversion is carried out in different research areas, the uncertainty of the inversion result is increased due to too few constraint conditions when 1 or a plurality of vegetation indexes are singly used for inversion, 2 geographic environment factors of elevation values and vegetation coverage are introduced to construct a multiple linear regression inversion model of the geographic environment factors, LAI sensitive wave bands and actually measured LAI, the precision of the LAI inversion model is improved, and the LAI inversion model can be widely applied to the departments of agriculture, environmental protection, water conservancy, China and the like.
Drawings
FIG. 1 is a schematic diagram of a basic process flow of an embodiment of the present invention.
Fig. 2 is a graph showing the annual changes of LAI in fuzhou in summer according to an embodiment of the present invention.
Fig. 3 is a spatial layout diagram of the LAI in 2018 summer in fuzhou city according to an embodiment of the present invention.
Fig. 4 is a graph of the stability level of LAI in summer in fuzhou city in 2008 + 2018 obtained in the embodiment of the present invention.
Fig. 5 is a characteristic graph of the LAI variation trend in summer in fuzhou city in 2008 + 2018 obtained in the embodiment of the present invention.
Fig. 6 is a graph of the sustainability characteristics of the LAI in summer in fuzhou city in 2008 + 2018 according to an embodiment of the present invention.
Fig. 7 is a trend graph of K according to altitude according to an embodiment of the present invention.
FIG. 8 is a graph showing the area ratio at different slopes for varying types of LAI made by an embodiment of the present invention.
Fig. 9 is a comparison of the K-slope variation characteristics (left) and the LAI variation types obtained by the embodiment of the present invention in the area ratio (right) of different slopes.
Fig. 10 is a LAI profile of different land cover types obtained by an embodiment of the present invention.
Detailed Description
As shown in the attached drawings, the implementation steps of the leaf area index space-time change characteristic research method of the embodiment are as follows:
1) acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, constructing an LAI inversion model, inverting LAI time series data of the region to be monitored, obtaining the LAI time series data of the region to be monitored, and carrying out precision verification;
2) performing space-time change analysis on the area to be monitored, after LAI time sequence data of the area to be monitored is obtained, analyzing the space volatility of the LAI of the area to be monitored by using a variation coefficient pixel by pixel, analyzing the space change trend of the LAI pixel by a unary linear regression analysis method, analyzing the sustainability of the change of the LAI pixel by an R/S analysis method, and analyzing the relation between the type of the change trend of the LAI and a terrain factor by combining terrain data;
3) carrying out land cover classification on remote sensing image data of an area to be monitored by using a supervision classification method, counting average pixel LAI values of various land cover types and benefits of land cover change on total LAI, introducing a GWR (global warming potential) model to analyze the relation between different land cover types and the LAI and the spatial non-stationarity of the different land cover types and the LAI so as to reveal the spatial non-stationarity relation between the land cover and the area to be monitored, and obtaining the relation between different land cover type patterns and the area to be monitored LAI;
as a further improvement of the technical scheme of the invention:
the method comprises the following steps of 1), acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, and constructing an LAI inversion model, wherein the detailed step of inverting LAI time sequence data of the region to be monitored comprises the following steps:
A1) when field actual measurement data of an area to be monitored is obtained, sampling and selecting a 30m multiplied by 30m sample plot for investigation, selecting a place with uniform vegetation distribution and representativeness as the sample plot, setting sampling points at four corners and the center of each sample plot, carrying out vegetation canopy leaf area index collection on each sampling point by using an LAI-2000 plant canopy analyzer, and finally taking the average value of the sampling points in a sample square as the LAI value of the sample square to obtain the LAI value of the sample square;
A2) when an inversion model is constructed, 2 geographic environment factors of vegetation coverage and elevation values are introduced as key factors of the inversion model, the key factors and sensitive bands of LAI are used as model input parameters together, a multiple linear regression model of the LAI sensitive bands, the geographic environment factors and actually measured LAI is constructed by using remote sensing image data and field sample data, and LAI time sequence data of an area to be monitored is obtained;
introducing vegetation coverage and 2 geographic environment factors with elevation values as key factors of an inversion model in the step A2), specifically, taking the vegetation coverage as an input parameter to participate in construction of the inversion model, inverting the vegetation coverage by using a pixel binary model based on NDVI (normalized difference of variance), and calculating the vegetation coverage according to the difference of the vegetation coverage and the elevation value
Figure 927461DEST_PATH_IMAGE001
Obtaining vegetation coverage, wherein NDVIsoil is the NDVI value of a region which is completely bare or not covered by vegetation, and NDVIveg is the NDVI value of a region which is completely covered by plants, and a method for determining NDVIsoil and NDVIveg is provided by the first plum seedling of the author;
in the step 2), the spatial volatility of the LAI of the area to be monitored is analyzed pixel by using the variation coefficient, the spatial variation trend of the LAI is analyzed pixel by a unary linear regression analysis method, and the detailed step of analyzing the sustainability of the LAI variation pixel by an R/S analysis method comprises the following steps:
B1) performing pixel-by-pixel stability analysis on LAI data of the area to be monitored, and evaluating the stability of the LAI along with the change of time by using the variation coefficientQualitative in terms of
Figure 671426DEST_PATH_IMAGE002
Obtaining the fluctuation condition of vegetation LAI of an area to be monitored, wherein n represents the number of years of monitoring, xi represents the annual LAI value, x represents the average value of the LAI, the larger CV is, the more scattered the data is, the smaller CV is, the more compact the data is, and the more stable the vegetation change is;
B2) quantitatively researching the change trend of vegetation LAI of the area to be monitored by adopting a unary linear regression analysis method pixel by pixel, wherein the obtained value is the slope of the unary linear regression analysis of each pixel in the monitoring year
Figure 33006DEST_PATH_IMAGE003
Obtaining the variation trend of the region to be monitored, wherein n is the total year of variation monitoring, Li is the value of LAI in the ith year, and SLOPE is the SLOPE of unary linear regression analysis;
B3) the sustainability of LAI change is analyzed pixel by adopting an R/S analysis method, the H value calculated by the R/S analysis method, namely the Hurst index can better reflect the autocorrelation of time sequence data, and the sustainability of the LAI change trend of vegetation in the area to be monitored pixel by pixel can be obtained by combining the Hurst index obtained by the analysis result of the R/S analysis method with the result of the change trend analysis;
the detailed step that the sustainability of the pixel-by-pixel variation trend of vegetation LAI of the area to be monitored can be obtained by combining the Hurst index obtained from the analysis result of the R/S analysis method in the step B3) with the variation trend analysis result comprises the following steps:
B31) according to
Figure 571435DEST_PATH_IMAGE004
Calculating a differential sequence of the LAI of the area to be monitored according to the differential sequence
Figure DEST_PATH_IMAGE013
Obtaining a mean sequence of the LAI;
B32) according to
Figure 487307DEST_PATH_IMAGE006
Is obtained to standMonitoring the dispersion of the area according to
Figure 34963DEST_PATH_IMAGE007
Obtaining the range of the region to be monitored, and finally obtaining the range according to
Figure 736203DEST_PATH_IMAGE014
Obtaining the standard deviation, the Hurst index is the ratio of R (m)/S (m), when 0<H<When the time sequence data of the LAI is 0.5, the trend change characteristics before and after the time sequence data of the LAI are opposite, namely the change of the LAI has reverse persistence, and the closer the H index is to 0, the stronger the reverse persistence is; when 0.5<H<When 1, the front trend and the back trend of the LAI time sequence data are consistent, namely the change of the LAI has sustainability, and the closer the H index is to 1, the stronger the sustainability is; when H =0.5, the change of the LAI time series data before and after is random, that is, the change trend cannot be judged;
analyzing the relationship between the LAI and the LAI variation trend type and the terrain factor in the step 2) by combining the terrain data, specifically, eliminating the terrain area difference by using a terrain area correction coefficient K to reveal the distribution and the evolution trend of the LAI variation type on different terrain features, according to the distribution and the evolution trend
Figure 976691DEST_PATH_IMAGE009
Obtaining a K value, wherein delta Vi is the area of a certain LAI change type under a certain topographic feature, and delta V is the area of a certain vegetation type in the whole research area; ti is the area of a particular topographical feature; s is the area of the whole study area;
the detailed step of introducing a GWR model in the step 3) to analyze the relationship between different land cover types and the LAI and the spatial non-stationarity of the land cover types and obtain the relationship between different land cover type patterns and the LAI of the area to be monitored comprises the following steps:
C1) the GWR model is based on the first law of geography, adds the geographic location of the data to the regression parameters of the model on the basis of an OLS (orientation left Square) model, taking into account the spatial weights of neighboring points, allowing local rather than global parameter estimation, based on which
Figure 379860DEST_PATH_IMAGE010
Obtaining a dependent variable of a sample point i, wherein yiIs a dependent variable, beta, of a sample point i0Is a constant term of regression, betak As regression parameter, xikIs the k-th variable observed for sample point i, εiFor error term, (mu)i，νi) Is the geographic coordinate, β, of the sample point i0（μi，νi) As the regression constant term, beta, for the sample point ik（μi，νi) Is xkRegression parameters at sample points i;
C2) according to
Figure DEST_PATH_IMAGE015
Calculating betak（μi，νi) Wherein X is a matrix of independent variables, XTIs the transpose of matrix X, Y is the dependent variable matrix, W (mu)i，νi) A matrix that is a spatial weight;
C3) according to
Figure 200048DEST_PATH_IMAGE016
Calculating a matrix of spatial weights, wherein WijTo estimate the weight of the point i to be measured using the spatial sample point j, dijAnd h is the bandwidth, and the Euclidean distance between the sampling point j and the point i to be estimated is determined by utilizing the minimum Chi information criterion.
The method is proved to be feasible by taking Fuzhou as a specific embodiment:
remote sensing image and preprocessing
The data used in the embodiment of the invention is Landsat series remote sensing image data with the resolution of 30m, and the download address is the official network of the United states geological exploration bureau (https:// earth explorer. usgs. gov /). The requirements for image selection are as follows: the time for obtaining the image is 8-9 months of the current year as much as possible, and in order to guarantee inversion accuracy, data without clouds or with few clouds are selected as much as possible. And for the years with a small amount of cloud coverage, generating a cloud mask file by using an ENVI cloud detection tool to perform cloud removing treatment, and inlaying data of other months in the current year to the removed area due to the cloud removing treatment. The measurement of the ground measurement data used was carried out using an LAI-2000 vegetation canopy analyzer manufactured by LI-COR, USA. The sampling dates were 8 and 9 months in summer of 2017, and in addition, in order to verify the accuracy of LAI data for other years of inversion using the inversion model, the previously measured LAI historical data of 2011 and 2016 were used. The historical data in 2011 and 2016 are used for verifying the accuracy of inversion of an inversion model in other years, 9 sample points are acquired in 2011, and 22 sample points are acquired in 2016, and are respectively used for verifying the inversion data in 2011 and 2016.
1 inversion model construction method
The method is characterized in that 60 groups of actually measured sample point data collected in the field are used for model construction and precision inspection, a statistical tool for model construction and a tool for precision inspection are completed by using SPSS, and the specific implementation method is as follows: sensitivity analysis is carried out on a GSA (Global Sensitivity analysis) tool kit, and a wave band with higher Sensitivity to vegetation LAI is selected. Secondly, introducing key geographic environment factors influencing vegetation distribution into the model as dependent variables of the constraint model, and constructing a multiple linear regression model of the geographic environment factors, the sensitive wave bands and the actually measured LAI.
1.1 sensitivity assay
The sensitivity analysis is to qualitatively or quantitatively analyze the sensitive band of the vegetation biochemical parameters, and before the LAI inversion is carried out, in order to determine the sensitive band of the LAI parameter and improve the inversion accuracy, the parameter sensitivity analysis is firstly carried out to find out the band sensitive to the LAI. The parameter sensitivity analysis is divided into two categories: local sensitivity analysis and global sensitivity analysis. The local analysis of parameter sensitivity is to analyze the influence of a single parameter on a model and set other parameters as fixed constants, and has the advantages of easy operation, neglecting the mutual influence among the parameters and adopting the analysis method in the previous sensitivity analysis. The global sensitivity analysis can make up for the defects of the qualitative analysis, so that the method for global sensitivity analysis is more and more widely applied in recent years. In order to improve the efficiency of global analysis, parameters which are obviously small in sensitivity in each wave band in local analysis are removed and do not participate in subsequent global sensitivity analysis. The global sensitivity analysis method mainly comprises a Sobol method, a FAST method, an EFAST method and a modified Sobol method. The first author, Saltelli, proposed in 2010 an improved Sobol algorithm, a variance-based global sensitivity analysis algorithm, for calculating first-order sensitivity and total sensitivity. Compared with the original algorithm, the improved Sobol algorithm has stronger stability and smaller calculation cost. The global Sensitivity analysis uses the gsa (global Sensitivity analysis) kit developed by the university of vienna, which can be run on Matlab.
1.2 acquisition of the geographic Environment factor
Vegetation coverage is the percentage of the total area of the projected area of the vegetation on the ground. The researchers such as Lirongchun and Luxioming find that the vegetation coverage and the LAI have a significant correlation, and the LAI of the corn and the rice is inverted by using the vegetation coverage and the actually measured LAI. Therefore, the research takes vegetation coverage as an input parameter to participate in the construction of the inversion model. Inversion of vegetation coverage using NDVI-based pixel dichotomy models, the formula is as follows:
Figure DEST_PATH_IMAGE017
wherein NDVIsoil is the NDVI value of an area completely covered by bare soil or no vegetation, and NDVIveg is the NDVI value of an area completely covered by vegetation, according to the method for determining NDVIsoil and NDVIveg proposed by the first author plum seedling.
From the above formula, in the same climate zone, the terrain is one of the important factors influencing the vegetation distribution. The learners like junk and peace successfully construct an LAI inversion model with higher precision than that without considering the influence of topographic factors on vegetation distribution. The Fuzhou terrain and landform are complex, so that the elevation value of the actually measured sample data is introduced to be used as a geographic environment factor to participate in the construction of an inversion model, and the terrain factor is extracted by using 30m DEM data.
1.3 stability analysis
In order to analyze the fluctuation situation of vegetation LAI in a research area for 11 years, the LAI data in the past 11 years are analyzed by pixel stability. The coefficient of variation is a statistic used to measure the degree of variation of a set of data, and is used to evaluate the stability of LAI over time, and the formula is:
Figure 473903DEST_PATH_IMAGE018
in the formula, n represents the number of years monitored and is 11 years, xi represents the LAI value year by year, and x represents the average value of LAI over 11 years. Larger CV values indicate more scattered data, smaller CV values indicate more compact data, and vegetation change is more stable. Stability analysis was performed using Band Math tool from ENVI.
1.4 analysis of Change trends
When grid analysis of a long-time sequence is carried out, the change trend of the LAI value of each pixel in the last years needs to be known, the change trend of vegetation LAI in Fuzhou city is quantitatively researched by adopting a unary linear regression analysis method pixel by pixel, the obtained value is the slope of the unary linear regression analysis of each pixel in the last 11 years, and the calculation formula is as follows:
Figure DEST_PATH_IMAGE019
where n is the total years of change monitoring 11, Li is the value of LAI in year i, and SLOPE is the SLOPE of the unary linear regression analysis. When SLOPE >0, LAI is increasing, and when SLOPE <0, LAI is decreasing. In general, only trend values that pass the significance test are reliable. And (4) carrying out pixel-by-pixel significance test on the result of the trend analysis to obtain a pixel-by-pixel significance P value. The trend of the change is divided into three types of non-significance (P > 0.05), significance (0.01 < P < 0.05) and extreme significance (P < 0.01). Trend analysis and significance testing were implemented using Matlab 2018 programming.
1.5R/S analysis
The R/S analysis method is also a weighing standard deviation analysis method, is a typing theory for quantitatively describing the time sequence information dependence, and the H value obtained by the R/S analysis method, namely the Hurst index, can better reflect the autocorrelation of time sequence data. The Hurst index obtained from the analysis result of the R/S analysis method is combined with the result of the change trend analysis, so that the sustainability of the LAI of the vegetation in Fuzhou city in the change trend from pixel to pixel can be obtained.
LAIi is a LAI time series, i =1, 2, 3 ·, n, which is defined for any positive integer m:
(1) difference sequence:
Figure 337823DEST_PATH_IMAGE020
(2) mean sequence:
Figure DEST_PATH_IMAGE021
(3) cumulative dispersion:
Figure 447861DEST_PATH_IMAGE022
(4) extremely poor:
Figure DEST_PATH_IMAGE023
(5) standard deviation:
Figure 55429DEST_PATH_IMAGE024
wherein the Hurst index is the ratio of R (m)/S (m). When 0< H <0.5, the trend change characteristics before and after the LAI time sequence data are opposite, namely the change of the LAI has reverse persistence, and the closer the H index is to 0, the stronger the reverse persistence is; when 0.5< H <1, the trend of LAI time sequence data is consistent, namely the change of LAI has sustainability, and the closer the H index is to 1, the stronger the sustainability is; when H =0.5, the change before and after the LAI time-series data is random, that is, the trend of the change cannot be judged.
1.6 correction of topographic area differences
The distribution characteristics of LAIs under different terrain conditions may be different, for example, when the area proportion of a certain type of LAI variation trend on the certain terrain condition is larger than that of the LAI variation type on the whole research area under the specific terrain condition, the distribution characteristics of the LAI variation type under the certain terrain condition may be evaluated unreasonably. Therefore, the influence of this effect is eliminated to reveal the distribution and the evolution trend of the LAI variation types on different topographic features. The research uses a terrain area correction coefficient K to eliminate the effect, the K value is a relative value, the K value is introduced to avoid unreasonable influence on an analysis result caused by the fact that when a certain LAI change type occupies a large proportion on certain terrain features but occupies a small proportion in the whole research area, and the influence is avoided
The calculation formula is as follows:
Figure 731261DEST_PATH_IMAGE025
in the formula, Δ Vi is the area of a certain LAI variation type under a certain topographic feature, the LAI variation type is vegetation invariant, vegetation increment or vegetation decrement, and the topographic feature is a certain range of elevation and gradient or a certain direction of slope; Δ V is the area of a certain vegetation type throughout the study area; ti is the area of a particular topographical feature; s is the area of the whole study area; k >1 indicates that the distribution of the LAI variation type on the topographic feature is dominant distribution; when K =1, the distribution of the LAI type on the topographic feature is equal to the distribution of the vegetation variation type in the whole area, and when K <1 is a smooth distribution on the topographic feature, the LAI type belongs to a non-dominant distribution under the topographic feature.
1.7 Fuzhou City summer LAI interpersonal variability characteristics
And analyzing annual change trend characteristics of LAI in Fuzhou summer by using annual average value of 30m LAI data in Fuzhou summer inverted in 2008-2018, as shown in FIG. 2. Through statistics of annual average LAI of summer data of Fuzhou city in each year, overall leaf area index changes of the LAI in 11 years in summer in the Fuzhou city are small, the trend of rising in fluctuation is shown, the formula of a fitted line is y =0.009x-15.587, and the annual average LAI in 2008-2018 summer in the Fuzhou city is increased by 0.009 through P <0.01 significance test. The reason for this increase may be that in recent years, although cities have been rapidly expanding and encroach on a large amount of vegetation, LAI has generally been on an increasing trend year by year due to increasing emphasis on ecological protection, such as increase in urban greenbelts, implementation of policies of returning to farm land, returning to grass, and increase in dominant species. The LAI of Fuzhou city in 2008-2018 has a variation range of 3.40-3.66, the average value is 3.51, the maximum value appears in 2017 and reaches 3.66, which exceeds the average value of 0.15, and the minimum value appears in 2010 and is 3.40 which is lower than the average value of 0.11.
Analysis of LAI spatiotemporal variation in summer of 2 Fuzhou city
LAI, one of the important parameters of vegetation canopy, has an indicative meaning to the growth status of vegetation and is an important input parameter for many geographic models and ecological processes. In recent years, LAI has been studied mostly for single vegetation types such as crops, grassland and woodland, but relatively few studies have been conducted on an urban scale. In order to know the time variation trend and the spatial distribution characteristics of the summer LAI in Fuzhou city, the variation intensity, the variation mode and the variation sustainability of the summer LAI in Fuzhou city are analyzed by a data statistical method pixel by pixel. The variation coefficient analyzes the fluctuation of the LAI in summer in Fuzhou city, the unitary linear regression analyzes the variation trend of the LAI in summer in Fuzhou city, and the Hurst index obtained by the R/S analysis analyzes the sustainability of the LAI variation in summer in Fuzhou city. The terrain and the landform of Fuzhou are complex, and the terrain is an important factor influencing the vegetation distribution, so that the distribution characteristics of the Fuzhou vegetation LAI on different terrain factors are analyzed, and the terrain area difference correction coefficient is introduced to analyze the distribution characteristics of different vegetation change types on different terrain factors.
2.1 Fuzhou City summer LAI spatial distribution characteristics
Drawing the summer LAI data of 30m in 2018 to obtain a summer LAI spatial distribution map in Fuzhou city, as shown in FIG. 3. As can be seen from fig. 3, the LAI in the summer of fuzhou city is low in the east coast and high in the west inland, and gradually decreases from the west mountain area to the east plain. The area with higher LAI is mainly distributed in mountainous and hilly areas, and the area with lower LAI is mainly distributed in eastern plain city areas, islands along the sea. This is because mountains and hills are less affected by human activities, cities with relatively high LAI are more affected by human activities, and LAI is relatively low.
2.2 Fuzhou City summer LAI stability analysis
And analyzing and calculating the stability of the Fuzhou summer LAI by using the Fuzhou summer LAI data obtained by inversion in 2008-2018 to obtain a variation coefficient spatial distribution map, as shown in FIG. 4. The variation coefficient reflects the intensity of change of vegetation LAI on a time sequence, the CV variation coefficient of the vegetation LAI in 2008-2018 of Fuzhou city is between 0 and 2.8 in summer, and the average value is 0.4. The coefficient of variation of summer vegetation LAI in Fuzhou city is divided into five grades, namely extremely low fluctuation (0-0.22), low fluctuation (0.22-0.31), medium fluctuation (0.31-0.49), high fluctuation (0.49-0.58) and extremely high fluctuation (0.58-2.73). The extremely-low fluctuation area and the medium fluctuation area respectively account for 11.07 percent and 22.78 percent of the total, the medium fluctuation area accounts for 43.08 percent of the total, a part of the areas are mainly distributed in inland hilly areas and mountain areas, the areas are less influenced by human activities, and the intensity of the fluctuation is less; the other part is mainly distributed in areas with stable land covering types in towns. The high and extremely high fluctuation regions respectively account for 10.09% and 12.99% of the whole. The regions with higher volatility are mainly distributed in town regions, and regions with higher volatility are distributed in different degrees in each district and county, such as the Guian region in the Henjiang county, the Minhou county, the Changle city and the Tan county.
2.3 analysis of LAI Change trends in Fuzhou City in summer
And calculating coefficients of unary linear regression analysis between 11 years in Fuzhou city by using summer LAI data of 30m in Fuzhou city inverted in 2008-2018 pixel by pixel, wherein the positive and negative of the coefficients represent that the change trend of the LAI of the pixels is increased or decreased. Analysis of LAI trend in Fuzhou city in summer over 11 years and significance test, the spatial distribution diagram is shown in FIG. 5. In general, the areas of decreased summer LAI in fuzhou city are slightly larger than the areas of increased vegetation LAI, which account for 44.05% and 55.95% of the entire province, respectively, but only 23.33% of the areas pass the significance test, indicating that the summer LAI in most areas of fuzhou city does not change significantly over the 11 years.
TABLE 1 LAI Trend variation grade statistical table for each county
Figure 313552DEST_PATH_IMAGE026
2.4 Fuzhou City summer LAI Change sustainability analysis
In order to analyze the sustainability of the LAI variation trend of the summer vegetation in Fuzhou city, the Hurst index distribution data of the LAI in summer in Fuzhou city from 2008 to 2018 is obtained through R/S analysis by using the 30m summer LAI data of the Fuzhou city from 2008 to 2018, and the variation trend analysis grade data is superposed, so that the sustainability of the LAI variation of the Fuzhou summer vegetation can be obtained, as shown in FIGS. 4 to 6. As can be seen from FIG. 6, the Hurst index varies from 0.06 to 1.06, with an average value of 0.52. According to the experience of Liu Yan preface of the first author, the value of the Hurt index between 0.4 and 0.6 is regarded as random change, the continuity of the change cannot be judged, and the area which cannot be judged whether the change has the sustainability accounts for 56.81%, which indicates that the future change trend cannot be judged by LAI in summer in most areas of Fuzhou city. When the Hurt index is less than 0.4, this indicates reverse persistence, 17.46% of the total area. When the Hurt index is greater than 0.6, sustainability is demonstrated, accounting for 25.73% of the total area. Overlapping the Hunt index and the variation trend, and dividing the Hunt index and the variation trend into 5 categories, wherein the continuously reduced area accounts for 14.05%; the region of continued increase is 11.68%, the region of change from increase to decrease is 6.17%, and the region of change from decrease to increase is 11.29% of the total area.
TABLE 2 LAI sustainability level statistics table for each county
Figure 956892DEST_PATH_IMAGE027
2.5 Distinguishing Effect of altitude on LAI changes in Fuzhou City in summer
The vegetation change types have larger difference in distribution at different altitudes, and as can be seen from fig. 7, with the rise of the altitude, the overall change trend of LAI in fuzhou in summer is that the LAI increases first and then decreases, the LAI does not change stably and increases, the LAI decreases first and then increases and then decreases, and the dominant distribution intervals of the different vegetation LAI change types are different. The LAI reduction type has a K value larger than the LAI invariant and the LAI reduction type when the elevation is less than 200m, which shows that the LAI reduction type is a dominant distribution type in the range of the elevation less than 200 m; in the range of elevation 200-500 m, the K value of the LAI unchanged is gradually increased, which shows that the LAI unchanged becomes an advantageous distribution type in the range of 200-500 m; in the range of 500-1000 m of elevation, the increasing K value of the LAI is gradually increased and becomes a dominant distribution type when being larger than the K values of the other two types in the range; in the range of elevation 1000-1400 m, the dominant distribution type is formed by the fact that the K of the unchanged LAI is larger than the other two types; within the range of 1400-1600 m, the K values of the LAI reduction type which are larger than the other two types become the dominant distribution type.
2.6 differential Effect of grade on LAI Change in summer in Fuzhou City
Fig. 8 shows the results of the gradient-dependent change in K values for the LAI variation types. As can be seen from fig. 8, the gradient has a small influence on the LAI invariant type, K of the invariant type is always about 1, and the influence on the LAI increase type and the LAI decrease type is large. When the gradient is less than 10 °, the LAI reduction type is the distribution dominance distribution type, but in this gradient range, the K value of the LAI reduction type gradually decreases, indicating that in this range, the distribution of the LAI reduction type tends to decrease as the gradient increases. Along with the increase of the gradient, the K value of the LAI reducing type is gradually increased, and the K value of the LAI increasing type is more than that of the other two types and presents advantageous distribution within the range of 10-30 degrees of the gradient. When the gradient is greater than 30 °, the K value of the LAI-invariant is greater than the other two types, presenting a dominant distribution.
2.7 differentiation of LAI changes in Fuzhou City in summer from the slope
As can be seen from fig. 9, flat ground has a large influence on the change of LAI, the distribution advantage of LAI decreasing is more obvious, the K value is 1.21, the LAI increasing type distribution is weaker, and K is 0.84, and the influence of human activities is large in the flat ground area, resulting in the LAI decreasing type being the dominant type. The influence of the change types of the LAI is small in all slopes except the flat ground, and the K values are all around 1. The occupation ratios of the land areas are reduced from the flat land and are larger than the occupation ratios of other slope directions, the occupation ratios of the land areas are increased from the flat land and are smaller than the occupation ratios of other slope directions, but the occupation ratio difference of the LAI unchanged type is smaller. In summary, the flat ground has a large influence on human activities, a large influence on the distribution of the LAI reduction and increase, and other slopes have a certain difference but a small influence on the distribution of various types of LAI changes.
Relationship between LAI and land cover in summer of 3 Fuzhou city
The types of land cover in fuzhou city have changed significantly due to the increased speed of economic development, population expansion and urbanization. Changes in land cover type can directly affect vegetation distribution and, in turn, the LAI of the vegetation. As a factor directly affecting vegetation LAI, it is necessary to analyze the response of vegetation LAI to changes in land cover. Firstly, land cover type data of 2008 and 2018 are acquired, and on the basis, LAI distribution characteristics of different land cover types are analyzed, and the loss and the benefit of vegetation LAI are reduced when the different land cover types are mutually converted. In previous researches, the influence degree of land cover is analyzed on the whole, so that the influence of different land cover types on the vegetation is ignored, and the influence of the land cover types on the vegetation is analyzed, so that the influence of the different land cover types on the vegetation is analyzed, and the influence of the different land cover types on the vegetation LAI at different spatial positions is analyzed.
3.1 Fuzhou City land cover classification
And (3) carrying out land coverage classification on the Landsat remote sensing images preprocessed in 2008 and 2018 by using a random forest method. After classification, the Google high-resolution data is used as auxiliary data to randomly select 50 verification samples for each land coverage type in the research area to perform precision verification on classification results, and a mixed matrix for precision verification is generated. The total accuracy of the two-stage land cover type data is about 80%, and the requirement of classification accuracy is met.
3.2 summer LAI Difference in different land cover types
In order to reveal the overall distribution characteristics of summer vegetation LAIs of different land cover types, summer LAI data and land cover data of 2008 and 2018 are superposed, the average value of LAIs of all pixels of different land cover types is counted, and fig. 10 is a histogram of LAI statistical data of different land cover types in fuzhou city. Overall, the summer average LAI of each land cover type was significantly different between 2008 and 2018 and showed consistent regularity. In 2008, the average of the LAI of different land coverage types in summer is forest land > farmland > grassland/shrub > land for construction > bare land, and in 2018, the average of the LAI of different land coverage types in summer is forest land > grassland/shrub > farmland > land for construction > bare land. The forest land summer LAI for the two-stage data is highest, 2.39 and 2.55 respectively. In time, the types of land whose LAI decreased in summer from 2008 to 2018 were farmlands and construction lands, and the summer LAI of woodland, grassland/shrub and bare land types were all increased.
3.3 summer LAI profit-loss analysis under land cover type conversion
Different land cover types have different ecological structures, and have different vegetation distributions and vegetation numbers. The conversion between different land cover types can directly influence the distribution and structure of vegetation, and further influence the size of unit LAI. The impact of land cover type transformation on vegetation LAI is mainly manifested in two aspects: one aspect is that the land cover type with higher vegetation LAI such as forest land, grassland and farmland is converted into the land cover type with lower vegetation LAI such as construction land; on the other hand, land cover types with low vegetation LAI in construction land, bare land and water are transformed into land cover types with higher vegetation LAI such as forest land, resulting in an increase in LAI.
TABLE 3 LAI profit-loss matrix under conversion of different land cover types
Figure 650042DEST_PATH_IMAGE028
3.4 unitary Linear model of land cover ratio Change and summer LAI
In the two-stage data, the proportion of the area of each land feature and the LAI in summer in Fuzhou city have a certain linear relationship, but the correlation degree is greatly different. In various land features, besides the positive correlation between the forest land and the LAI, the negative correlation is between the farmland, the construction land, the grassland/shrub and the bare land. The unary linear fitting equation shows that linear equations of various land types are in linear positive correlation or negative correlation through significance test P <0.01, and therefore the various land coverage types all contribute to the increase or decrease of the LAI. Among the land cover types, the land cover type which is positively correlated and has the largest correlation degree is forest land, and R2 in 2008 and 2018 is 0.618 and 0.786 respectively; the land cover types which are in negative correlation and have the maximum correlation degree are construction lands, R2 of 2008 and 2018 are respectively 0.555 and 0.681, and other land feature types are in negative correlation with the LAI, but R2 is obviously smaller than the construction lands, so that the contribution rate of the construction lands to the LAI is most obvious in the land cover types which have negative effects on the LAI.
3.5 GWR model of land cover ratio Change to summer LAI
When the relation between the land cover proportion and the LAI is analyzed, in order to reveal the spatial non-stationary relation between the LAI and different land cover types, a geographical Weighted regression model GWR (geographic Weighted regression) capable of reflecting the relation characteristics between local geographic elements is used. The GWR model is based on the first law of geography, and on the basis of the ols (geographic Least square) model, adds the geographic location of the data to the regression parameters of the model, and allows local parameter estimation instead of global estimation, taking into account the spatial weights of neighboring points, and the formula is:
Figure DEST_PATH_IMAGE029
in the formula, yiIs a dependent variable, beta, of a sample point i0Is a constant term of regression, betakIs a regression parameter, X, of the variable kikIs the k-th variable observed for sample point i, εiFor error term, (mu)i，νi) Is the geographic coordinate, β, of the sample point i0（μi，νi) Is the regression constant term, β, of sample ik（μi，νi) Is XkRegression parameters at sample points i. Beta is ak（μi，νi) The formula for the estimate of (c) is as follows:
Figure 383642DEST_PATH_IMAGE030
in the formula: x is a matrix of independent variables, XTIs the transpose of matrix X, Y is the dependent variable matrix, W (mu)i，νi) Is a matrix of spatial weights. The spatial weight matrix is an estimate of betak（μi，νi) Is realized by adopting a Gaussian function, and the formula is shown as follows
Figure DEST_PATH_IMAGE031
In the formula WijTo estimate the weight of the point i to be measured using the spatial sample point j, dijThe Euclidean distance between a sampling point j and a point i to be estimated is taken as h, and bandwidth is taken as h, and the minimum Red pool information criterion (AIC) is used for determining. The GWR model was implemented using ArcGIS10.3 software.
The results of the GWR model fitting for each land cover type versus summer LAI are shown in table 4. On the whole, the fitting result of the GWR model is consistent with that of a unitary linear regression model, the average value of regression coefficients of farmlands, grasslands/shrubs, construction land and bare land is less than 0, the farmlands, the grasslands/shrubs and the construction land mainly comprise regions less than 0, all the regions of the bare land are less than 0, and the 4 land coverage types mainly reduce the LAI; the mean value of the forest land is greater than 0, the regression coefficient is greater than 0 in all areas, and the effect of the forest land on the research area is mainly used for increasing the LAI. However, the influence of the five types of land cover on the LAI of the research area has obvious spatial non-stationarity, and the vegetation LAI has large difference at different spatial positions. The standard deviation of the regression coefficients in 2008 and 2018 is the bare land with the largest standard deviation and the forest land with the smallest standard deviation. The regression coefficient of the bare land is proved to have larger change at different spatial positions in the research area, and the spatial non-stationarity of the bare land is strongest in the five-land coverage type; the change of the woodland in different space positions is small, and the space instability of the woodland is the weakest in the five-land covering type.
TABLE 4 GWR model parameters for land cover vs LAI
Figure 612802DEST_PATH_IMAGE032
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment, and all technical solutions belonging to the principle of the present invention belong to the protection scope of the present invention. It will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.

Claims (4)

1. A method for researching the space-time change characteristics of a leaf area index is characterized by comprising the following steps:
1) acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, constructing an LAI inversion model, inverting LAI time series data of the region to be monitored, obtaining the LAI time series data of the region to be monitored, and carrying out precision verification;
2) performing space-time change analysis on the area to be monitored, after LAI time sequence data of the area to be monitored is obtained, analyzing the space volatility of the LAI of the area to be monitored by using a variation coefficient pixel by pixel, analyzing the space change trend of the LAI pixel by a unary linear regression analysis method, analyzing the sustainability of the change of the LAI pixel by an R/S analysis method, and analyzing the relation between the type of the change trend of the LAI and a terrain factor by combining terrain data;
3) the method comprises the steps of carrying out land cover classification on remote sensing image data of an area to be monitored by using a supervision classification method, counting average pixel LAI values of various land cover types and losses and benefits of land cover change to total LAI, introducing a GWR (global warming potential) model to analyze the relation between different land cover types and the LAI and the spatial non-stationarity of the different land cover types and the LAI so as to reveal the spatial non-stationarity relation between the land cover and the area to be monitored, and obtaining the relation between different land cover type patterns and the area to be monitored LAI.
2. The method for studying spatial-temporal variation characteristics of leaf area index according to claim 1, wherein: the method comprises the following steps of 1), acquiring a remote sensing image and field actual measurement data of a region to be monitored, carrying out radiometric calibration and atmospheric correction on an original remote sensing image, and constructing an LAI inversion model, wherein the detailed step of inverting LAI time sequence data of the region to be monitored comprises the following steps:
A1) when field actual measurement data of an area to be monitored is obtained, sampling and selecting a 30m multiplied by 30m sample plot for investigation, selecting a place with uniform vegetation distribution and representativeness as the sample plot, setting sampling points at four corners and the center of each sample plot, carrying out vegetation canopy leaf area index collection on each sampling point by using an LAI-2000 plant canopy analyzer, and finally taking the average value of the sampling points in a sample square as the LAI value of the sample square to obtain the LAI value of the sample square;
A2) when an inversion model is constructed, 2 geographic environment factors of vegetation coverage and elevation values are introduced as key factors of the inversion model, the key factors and sensitive bands of LAI are used as model input parameters together, a multiple linear regression model of the LAI sensitive bands, the geographic environment factors and actually measured LAI is constructed by using remote sensing image data and field sample data, and LAI time sequence data of an area to be monitored is obtained;
introducing vegetation coverage and 2 geographic environment factors with elevation values as key factors of an inversion model in the step A2), specifically, taking the vegetation coverage as an input parameter to participate in construction of the inversion model, inverting the vegetation coverage by using a pixel binary model based on NDVI (normalized difference of variance), and calculating the vegetation coverage according to the difference of the vegetation coverage and the elevation value
Figure 433564DEST_PATH_IMAGE001
Obtaining vegetation coverage, wherein NDVIsoil is the NDVI value of the area completely covered by bare soil or no vegetation, and NDVIveg is the NDVI value of the area completely covered by plants, and adopting the method for determining NDVIsoil and NDVIveg proposed by the first author plum seedling.
3. The method for studying spatial-temporal variation characteristics of leaf area index according to claim 1, wherein: in the step 2), the spatial volatility of the LAI of the area to be monitored is analyzed pixel by using the variation coefficient, the spatial variation trend of the LAI is analyzed pixel by a unary linear regression analysis method, and the detailed step of analyzing the sustainability of the LAI variation pixel by an R/S analysis method comprises the following steps:
B1) performing pixel-by-pixel stability analysis on LAI data of the region to be monitored, and evaluating the stability of LAI along with time by using the variation coefficient according to
Figure 629053DEST_PATH_IMAGE002
Obtaining the fluctuation condition of vegetation LAI of an area to be monitored, wherein n represents the number of years of monitoring, xi represents the annual LAI value, x represents the average value of the LAI, the larger CV is, the more scattered the data is, the smaller CV is, the more compact the data is, and the more stable the vegetation change is;
B2) quantitatively researching the change trend of vegetation LAI of the area to be monitored by adopting a unary linear regression analysis method pixel by pixel, wherein the obtained value is the slope of the unary linear regression analysis of each pixel in the monitoring year
Figure 996580DEST_PATH_IMAGE003
Obtaining the variation trend of the region to be monitored, wherein n is the total year of variation monitoring, Li is the value of LAI in the ith year, and SLOPE is the SLOPE of unary linear regression analysis;
B3) the sustainability of LAI change is analyzed pixel by adopting an R/S analysis method, the H value calculated by the R/S analysis method, namely the Hurst index can better reflect the autocorrelation of time sequence data, and the sustainability of the LAI change trend of vegetation in the area to be monitored pixel by pixel can be obtained by combining the Hurst index obtained by the analysis result of the R/S analysis method with the result of the change trend analysis;
the detailed step that the sustainability of the pixel-by-pixel variation trend of vegetation LAI of the area to be monitored can be obtained by combining the Hurst index obtained from the analysis result of the R/S analysis method in the step B3) with the variation trend analysis result comprises the following steps:
B31) according to
Figure 175889DEST_PATH_IMAGE004
Calculating a differential sequence of the LAI of the area to be monitored according to the differential sequence
Figure 388696DEST_PATH_IMAGE005
Obtaining a mean sequence of the LAI;
B32) according to
Figure 704271DEST_PATH_IMAGE006
Obtaining the dispersion of the area to be monitored according to
Figure 977120DEST_PATH_IMAGE007
Obtaining the range of the region to be monitored, and finally obtaining the range according to
Figure 378146DEST_PATH_IMAGE008
Obtaining the standard deviation, the Hurst index is the ratio of R (m)/S (m), when 0<H<When the time sequence data of the LAI is 0.5, the trend change characteristics before and after the time sequence data of the LAI are opposite, namely the change of the LAI has reverse persistence, and the closer the H index is to 0, the stronger the reverse persistence is; when 0.5<H<When 1, the front trend and the back trend of the LAI time sequence data are consistent, namely the change of the LAI has sustainability, and the closer the H index is to 1, the stronger the sustainability is; when H =0.5, the change of the LAI time series data before and after is random, that is, the change trend cannot be judged;
analyzing the relationship between the LAI and the LAI variation trend type and the terrain factor in the step 2) by combining the terrain data, specifically, eliminating the terrain area difference by using a terrain area correction coefficient K to reveal the distribution and the evolution trend of the LAI variation type on different terrain features, according to the distribution and the evolution trend
Figure 660222DEST_PATH_IMAGE009
Obtaining a K value, wherein delta Vi is the area of a certain LAI change type under a certain topographic feature, and delta V is the area of a certain vegetation type in the whole research area; ti is the area of a particular topographical feature; s is the area of the entire study area.
4. The method for studying spatial-temporal variation characteristics of leaf area index according to claim 1, wherein: the detailed step of introducing a GWR model in the step 3) to analyze the relationship between different land cover types and the LAI and the spatial non-stationarity of the land cover types and obtain the relationship between different land cover type patterns and the LAI of the area to be monitored comprises the following steps:
C1) the GWR model is based on the first law of geography, adds the geographic location of the data to the regression parameters of the model on the basis of an OLS (orientation left Square) model, taking into account the spatial weights of neighboring points, allowing local rather than global parameter estimation, based on which
Figure 95883DEST_PATH_IMAGE010
Obtaining a dependent variable of a sample point i, wherein yi is the dependent variable of the sample point i, β 0 is a regression constant term, β k is a regression parameter, xik is a k-th variable observed value of the sample point i, ε i is an error term, (μ i, ν i) is a geographic coordinate of the sample point i, β 0 (μ i, ν i) is a regression constant term of the sample point i, and β k (μ i, ν i) is a regression parameter of xk at the sample point i;
C2) according to
Figure 805213DEST_PATH_IMAGE011
Calculating beta k (mu i, vi i), wherein X is a matrix of independent variables, XT is a transposed matrix of the matrix X, Y is a matrix of dependent variables, and W (mu i, vi) is a matrix of space weight;
C3) according to
Figure 693534DEST_PATH_IMAGE012
And calculating a space weight matrix, wherein Wij is the weight when a space sampling point j is used for estimating a point i to be measured, dij is the Euclidean distance between the sampling point j and the point i to be estimated, and h is the bandwidth, and the minimum Chi information criterion is used for determining.
CN202110731055.XA 2021-06-30 2021-06-30 Leaf area index space-time change characteristic research method Active CN113340898B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110731055.XA CN113340898B (en) 2021-06-30 2021-06-30 Leaf area index space-time change characteristic research method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110731055.XA CN113340898B (en) 2021-06-30 2021-06-30 Leaf area index space-time change characteristic research method

Publications (2)

Publication Number Publication Date
CN113340898A true CN113340898A (en) 2021-09-03
CN113340898B CN113340898B (en) 2023-05-30

Family

ID=77481554

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110731055.XA Active CN113340898B (en) 2021-06-30 2021-06-30 Leaf area index space-time change characteristic research method

Country Status (1)

Country Link
CN (1) CN113340898B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116720624A (en) * 2023-06-16 2023-09-08 西北农林科技大学 River basin vegetation evapotranspiration prediction method based on remote sensing data

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105303063A (en) * 2015-12-03 2016-02-03 武汉大学 Leaf area index inversion method and system of merged phenological data and remote sensing data
CN108229403A (en) * 2018-01-08 2018-06-29 中国科学院遥感与数字地球研究所 A kind of mixed model construction method for being used to estimate vegetation leaf area index
WO2018153143A1 (en) * 2017-02-22 2018-08-30 河海大学 Method for measuring mudflat elevation by remotely sensed water content
CN109886962A (en) * 2019-03-27 2019-06-14 南京林业大学 The remote sensing inversion method of vegetative coverage and the management measure factor based on LAI and data of multiple angles
CN111126511A (en) * 2020-01-02 2020-05-08 国智恒北斗好年景农业科技有限公司 Vegetation index fusion-based LAI quantitative model establishment method
GB202101106D0 (en) * 2020-07-14 2021-03-10 Aerospace Information Research Institute Chinese Academy Of Sciences Method and device for performing inversion of crop leaf area index

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105303063A (en) * 2015-12-03 2016-02-03 武汉大学 Leaf area index inversion method and system of merged phenological data and remote sensing data
WO2018153143A1 (en) * 2017-02-22 2018-08-30 河海大学 Method for measuring mudflat elevation by remotely sensed water content
CN108229403A (en) * 2018-01-08 2018-06-29 中国科学院遥感与数字地球研究所 A kind of mixed model construction method for being used to estimate vegetation leaf area index
CN109886962A (en) * 2019-03-27 2019-06-14 南京林业大学 The remote sensing inversion method of vegetative coverage and the management measure factor based on LAI and data of multiple angles
CN111126511A (en) * 2020-01-02 2020-05-08 国智恒北斗好年景农业科技有限公司 Vegetation index fusion-based LAI quantitative model establishment method
GB202101106D0 (en) * 2020-07-14 2021-03-10 Aerospace Information Research Institute Chinese Academy Of Sciences Method and device for performing inversion of crop leaf area index

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
YONGHUA QU等: "Simultaneous measurements of corn leaf area index and mean tilt angle from multi-directional sunlit and shaded fractions using downward-looking photography", COMPUTERS AND ELECTRONICS IN AGRICULTURE *
严恩萍;林辉;洪奕丰;张雨;陈利;: "杉木人工林叶面积指数估测及影响因子分析", 水土保持研究 *
帅晨;沙晋明;林金煌;季建万;周正龙;高尚;: "不同下垫面遥感指数与地温关系的空间差异性研究", 地球信息科学学报 *
蔡雯洁等: "基于地理环境要素的叶面积指数遥感定量反演", 亚热带资源与环境学报 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116720624A (en) * 2023-06-16 2023-09-08 西北农林科技大学 River basin vegetation evapotranspiration prediction method based on remote sensing data

Also Published As

Publication number Publication date
CN113340898B (en) 2023-05-30

Similar Documents

Publication Publication Date Title
Leroux et al. Crop monitoring using vegetation and thermal indices for yield estimates: case study of a rainfed cereal in semi-arid West Africa
CN111727443A (en) Soil available nutrient inversion method based on deep neural network
CN102073869A (en) Point location data and remote sensing image data-based regional farmland quality monitoring method
CN108984803A (en) A kind of method and system of crop yield spatialization
He et al. Mapping regional evapotranspiration in cloudy skies via variational assimilation of all-weather land surface temperature observations
Parida et al. Detecting drought-prone areas of rice agriculture using a MODIS-derived soil moisture index
CN109711102A (en) A kind of crop casualty loss fast evaluation method
CN112903606B (en) Mangrove forest ecological restoration force assessment method based on unmanned aerial vehicle hyperspectrum
CHENG et al. Cold damage risk assessment of double cropping rice in Hunan, China
CN114863289B (en) Dynamic remote sensing monitoring method and system based on land utilization
CN112819227A (en) County scale winter wheat yield per unit prediction method and system
Park et al. Integrated model for predicting rice yield with climate change
CN113567358B (en) Cultivated land pollution monitoring and early warning system based on remote sensing and Internet of things
Bolun et al. Estimating rice paddy areas in China using multi-temporal cloud-free normalized difference vegetation index (NDVI) imagery based on change detection
Cheng et al. Mapping paddy rice yield in Zhejiang Province using MODIS spectral index
CN113340898A (en) Leaf area index space-time change characteristic research method
Chu et al. Mapping and forecasting of rice cropping systems in central China using multiple data sources and phenology-based time-series similarity measurement
AU2021105767A4 (en) Risk Assessment Method of Winter Wheat Summer Maize Disaster
CN115830464A (en) Plateau mountain agricultural greenhouse automatic extraction method based on multi-source data
CN114611699A (en) Soil moisture downscaling method and device, electronic equipment and storage medium
Sun et al. Comparison and evaluation of remote sensing indices for agricultural drought monitoring over Kazakhstan
Jahan Vegetation indices, FAPAR and spatial seasonality analysis of crops in southern Sweden
CN117556695B (en) Crop root soil water content simulation method based on deep learning
Wang et al. Comprehensive evaluation of soil moisture in farmland based on Ensemble Empirical Mode Decomposition and Partial Least Squares Regression
Qiao et al. Observation and Geoinformation

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
GR01 Patent grant
GR01 Patent grant