Remote Monitoring Of Crop Pigment Content And Its Vertical Distribution Within Canopies

Chlorophyll and carotenoid are two main and important photosynthetic pigments in the leaves of crops,which can indicate their photosynthetic capacity and growth condition.Study on the monitoring of dynamic changes of crop pigment content and its vertical distribution within canopies would play important roles in the growth condition monitoring,stresses diagnosis,fertilization decision making,and it would provide scientific support for agricultural management and pests and diseases monitoring.Remote sensing technology provides an effective tool in the estimation of canopy pigment content and its vertical distribution within crops'canopies.However,some of underlying mechanisms are not adequately understood.For instance,spectral indices tend to saturate with the high carotenoid content,soil background reflectance and different plant cultivars have influences on the estimation of carotenoid content and reduce the estimation accuracy,some information on the retrieval mechanism of the vertical distribution of leaf chlorophyll content are not available,the retrieval algorithms of leaf chlorophyll content based on the newly launched remote sensing data needs to be developed,etc..In this study,winter wheat were taken as the main target,the following four parts research work have been conducted,including remote estimation of canopy carotenoid content based on proximal hyperspectral and in situ remote sensing data,remote estimation of carotenoid content based on multi-angle hyperspectral remote sensing data,remote monitoring of vertical distribution of leaf chlorophyll content within canopies based on multi-angle hyperspectral data,and remote estimation of leaf chlorophyll content based on Sentinel-2 satellite remote sensing data.The following results were obtained:(1)In order to estimate chlorophyll content using remote sensing data,many spectral indices have been proposed in many literatures.However,they have proven to saturate with high carotenoid content.In this study,CTRI(Carotenoid triangle ratio index)was proposed based on in situ hyperspectral data,and the PROSPECT5+SAIL radiative transfer model was used for saturation analysis of spectral indices,the results indicated that compared to the other spectral indices tested,the new index CTRI had a good linear relationship with canopy carotenoid content,and it didn't show a saturation effect.It can reflect the whole dynamic changes of carotenoid content.(2)This study explored the application potential of hyperspectral data obtained from multi-angle directions in canopy carotenoid content estimation for crops.Compared to the nadir direction,spectral reflectance obtained from 20°to 40°backscattering directions improved the estimation accuracy of carotenoid content.Spectral features(i.e.absorption depth and absorption area)centered around 500 nm derived from spectral data obtained from 30?backscattering direction was found to greatly reduce the influences of soil background and different plant cultivars.It was the most suitable for winter wheat canopy carotenoid content estimation,with a coefficient of determination of 0.79 and a root mean square error of 19.03 mg/m~2.(3)Canopy spectral reflectance measured from different multi-angle observations can be used to detect the vertical distribution of leaf chlorophyll content effectively.However,the adequate information on the penetration characteristics inside the canopy of the light in the chlorophyll absorption spectral regions and spectral index formulas for vertical leaf chlorophyll content estimation are not yet available.In this study,wheat canopies were divided into three vertical layers according to different leaf positions,all possible two-band combinations from 400 to1000 nm in SR and NDVI types of indices,and all possible three-band combinations in CI(chlorophyll index)type of indices,were calculated.In this study,we systematically analyzed the correlations between leaf chlorophyll content and all possible combinations over spectral bands in the above three types of indices as well as published spectral indices at all viewing angles for each vertical layer,the optimal viewing angle for each vertical layer was found.In addition,leaf stacking effect and the penetration characteristic of the light inside the canopy for different chlorophyll absorption regions of the spectrum were taken into account,our results showed the most sensitive spectral band combinations for vertical leaf chlorophyll content estimation,i.e.the combinations of short green band(center around 520 nm)with NIR bands were efficient in estimating upper leaf chlorophyll content,whereas the red edge(center around 695 nm)paired with NIR bands were dominant in quantifying leaf chlorophyll in the lower layers.Our results provide the support for the selection of spectral wavebands and viewing angles to design ground-based and airborne sensors.(4)We studied the algorithms for leaf chlorophyll content retrieval using Sentinel-2 satellite data.Several machine learning algorithms(i.e.kernel ridge regression algorithm,Gaussian process regression,random forest algorithm and partial least square regression)were used for estimating crop leaf chlorophyll content at the regional scale,as well as compared to the results based on the method of vegetation indices.We mapped the multi-temporal leaf chlorophyll content distribution over the study region based on Gaussian process regression algorithm,which showed the dynamic changes of leaf chlorophyll content in different growth stages of winter wheat.In addition,the result indicated that the red edge bands of Sentinel-2 satellite can improve the accuracy of leaf chlorophyll content estimation.The main innovative contributions of this study are as follows:(1)In order to solve some of problems existed in the remote estimation of canopy carotenoid content,a new spectral index CTRI was proposed,which is highly sensitive to carotenoid content and strongly insensitive to the saturation effect.We explored the application potential of multi-angle hyperspectral data in crop canopy carotenoid content retrieval.The spectral features with directional information were developed,which can greatly reduce the influence of soil background and different crop plant cultivars,and therefore,improving the accuracy of carotenoid content estimation.(2)Several mechanisms of vertical leaf chlorophyll content estimation using multi-angle remote sensing technology were explored.By taking into account the factors of the penetration characteristics inside the canopy of the light in the chlorophyll absorption spectral regions and spectral index formulas,the most sensitive viewing angles,the optimal spectral band combinations and spectral index formulas were acquired for leaf chlorophyll content in each vertical layer.(3)We investigated retrieval algorithms for leaf chlorophyll content based on the characteristics of newly launched Sentinel-2 satellite data,and validated the important role of its red edge bands in the retrieval.This study provides insights in crop growth condition indicators monitoring for precision agriculture based on Sentinel-2 satellite data.