Remote Estimation Of Nitrogen Nutrition And Biophsyical Characteristics Of Summer Corn In Guanzhong Plain

The traditional method of measuring crop N status depended on plants sampled fromfields and chemical assays in laboratories, which could not meet the need of rapid, real-timeand nondestructive monitoring and effective diagnosis of plant N status. Remote monitoringof crops and the environment are agriculture (agricultural ecology) and informationtechnology, a combination of emerging cross-cutting areas of research. The newly emergedhyperspectral remote sensing makes possible in many narrow and continuous spectral bands,which are sensitive to specific crop variables and so weak difference in plant parameters,could be detected. Hyper-spectral remote sensing technology can be used to better estimatevarious growth variables related to crop physiology and biochemistry (plant growth, nutritionstatus of the crop, water stress, disease). The main objective of this study was used differenttechniques for remote monitoring of crop nutrition in real time which is most recent demandof time for precision agriculture and more over to develop system for directly control offertilizer application and better management of plant nutrition both economically by reducingtotal expenditure on fertilizer and to make environment friendly by reducing the potentialharmful runoff of fertilizers into stream and waterways. Another objective of this study was tomeasured corn biophysical characteristics (chlorophyll, LAI) non-destructively, develops newalgorithms for their precise estimation, will help in better understanding of the hyperspectralreflectance signature, and will provide better decision making in managing the crops.1). Hyperspectral remote sensing can improve accuracy and precision of estimation ofnitrogen contents in crops and leads effective management of nitrogen application in precisionagriculture. The objectives of this study were to identify spectral wavelengths, theircombinations and spectral vegetation indices (SVIs) that are indicative of nitrogen nutritionalcondition and to analyze the accuracy of different spectral parameters for remote estimation ofnitrogen status temporally. A field study was conducted during2010and2011at Northwest A&F University, China, to determine the relationship between leaf hyperspectral reflectance(350-1075nm) and leaf N contents in the field-grown corn (Zea mays L) under five nitrogenrates (0,60,120,180and240kg/ha pure nitrogen) were measured at key developmentalstages. The accuracy of nitrogen nutrition diagnosis among the single (R) and dual (R1+R2)waveband spectral reflectance, the algorithm of single (LgR) and dual (LgR1+LgR2)waveband spectral reflectance, spectral ratio (SR) in the green (NIR/G), red (NIR/R) and infrared (NIR/NIR), NDVI, GNDVI, and SAVI were compared. We employed linear andnon-linear model and chose the highest determination of coefficient (R~2) model and lowestRMSE and RRMSE at each stage, and recommended best model at each growth stage of thecorn crop. The results showed that there was the best fitting model between nitrogen contentsand single and dual spectral reflectance at R450, R550, R630, R680, R710and R550+R710at10-12leaf, followed by silking, tasseling, and late dent stage. The algorithm overall increasedthe accuracy both in the single and dual waveband. The fittings of the linear regression modelconstructed by spectra variables (LgR550+LgR720), Lg(R550+R710) and nitrogen contentswere the best among them. Spectral ratios in the NIR/Red with R810/R670showed highest R2and lowest RMSE at10-12leaf and silking stage followed by NIR/Green with R810/R550and NIR/NIR with R780/R700and at silking stage and10-12leaf stages respectively. GNDVIshowed the highest R2and lowest RMSE among all vegetation indices at10-12leaf stage.The results showed that Y=4.450+0.00x-17.99x~2+10.496x~3was the best prediction model forremote estimation of leaf N contents at10-12leaf stage followed by Y=-0.187-7.932X–3.452x~2, Y=3.092+1.684x+1.995x~2, Y=3.4345e-0.0113xand Y=0.964x-0.293at silking stage,tasseling stage, late dent stage and6-8leaf stage respectively. From the current study showedthat GNDVI, LogR550, R630, R810/R670and R680were good indicators for precisedetermining leaf N contents at different growth stages of the field grown summer corn. Thestudy results indicated that hyperspectral reflectance and spectral vegetation indices can beeffectively used as nondestructive, quick, reliable and in real time monitoring of corn nitrogenstatus and important tool for N fertilizer management in precision agriculture.2). The study employed another approach for remote monitoring of the crop nutritionstatus by using derivative spectroscopy (derivative reflectance spectra and their parameters)and red edge inflection point (REIP). The study comprised different nitrogen fertilizer ratesapplication along with two summer corn cultivars, crop reflectance spectrum and leaf totalnitrogen content were measured at different growth stages. The wavelength of470,550,620,720nm of crop reflectance of two corn cultivars were selected to assess regressionrelationship between leaf total nitrogen content and leaf spectral parameters, which includeoriginal reflectance, first order differential transform, derivative parameters (based onspectrum position, area, characteristic parameters of the vegetation index) and red edgeinflection point (REIP). Three high coefficients of determination (R~2) and F value modelswere chose at each growth stages to verify root mean square error (RMSE) and relative error(RE) for each cultivar for its self and inter-species cross-examination. Chose the smallest rootmean square error and relative error model was taken as the best model at each growth stages.The results showed that, the best fitting regression relationship between leaf total nitrogen content and spectrum parameter of R720, DR720, SDb, DR550and DR550were6-8leaf,10-12leaf, tasseling, silking and late dent stages of summer corn respectively. REIP showedhighest determination of coefficient (R2) at silking,10-12leaf, late dent, tasseling and6-8leafstages with the values of0.90,0.86,0.87,0.76, and0.71respectively. The study showed thatboth derivative reflectance spectra and red edge inflection point were better techniques foradequately, non-destructively and in real time estimation of leaf nitrogen contents at differentgrowth stages of summer corn.3). Leaf chlorophyll provides valuable information about the physiological status ofplants. Hyperspectral remote sensing makes it possible to assess plant information with moreprecisely, quickly and non-destructively. Recent studies have proved the practicability ofretrieval of chlorophyll content from hyperspectral vegetation indices composed by thereflectance of specific bands. The objective of this study was to investigate the spectralbehavior of the relationship between corn leaf reflectance and leaf chlorophyll content and todevelop a more suitable hyperspectral vegetation index for non-destructively estimation ofleaf chlorophyll content in summer corn. A field study was performed during2010-2011atAgricultural experimentation of Agronomy, Northwest A&F University, China, under fivenitrogen fertilizer rates (0,60,120,180and240kg/ha pure nitrogen). The leaf chlorophyllwas estimated at different growth stages of the corn during the growing season. In this studywe evaluated fourteen different hyperspectral vegetation indices including, NIR/R, NIR/G,NIR/NIR, CIred edge, CIgreen, CARI, MCARI, TCARI, NDVI, GNDVI, SAVI, and newimproved indices NIR/R, CIgreen II, CIred edge II, GNDVI/SAVI,. Results showed that differentindices were better for estimation of leaf chlorophyll content at different growth stages butnone of the indices was suitable for accurate estimation of the leaf chlorophyll content at allstages. The improve newly indices improved the chlorophyll estimation accuracy overall at allgrowth stages with R2ranges from0.66to0.81. Further, GNDVI/SAVI was validated againstgrain yield and showed strong relationship with the corn grain yield with R2ranges from0.528to0.828. The study results showed that hyperspectral spectral indices are more reliablefor quickly and nondestructively estimation of corn leaf chlorophyll contents and grain yield.This improved newly indices need further testing for its broad application against differentcrops and at different location and in analyzing digital airborne or satellite imagery to assist infertilizer management decision making.4). LAI is an important crop biophysical parameter for determining plant growth andyield. Hyperspectral vegetation indices proved that they are sensitive to vegetationbiophysical characteristics of the crops. The objective of this study was to evaluate the exitingspectral indices and to develop new algorithms that are more resistance to background crop reflectance for adequately prediction of green LAI. The study was performed during2010and2011at NWSUAF, Agricultural Research Station, Yangling, China, under five N rates tocompare the performances of existing vegetation indices. In this study, set of vegetationindices belonged to different classes (Normalized difference vegetation index (NDVI,GNDVI), simple ratio index (SRI), chlorophyll absorption ratio index (CARI, MCARI,TCARI), soil background resistance index (SAVI, MSAVI, OSAVI) were tested under fieldcondition to explore their potentials in estimation of LAI. Different bands combinations werealso used to develop the new improve modified vegetation indices in the integrated form(TCARI/MSAVI),(MCARI/MSAVI). The results showed that the above existing vegetationindices performed better at different growth stages but none of them were not appropriate atall growth stages because most of the indices were affected by saturation at high LAI. Thenewly improve spectral indices (TCARI/MSAVI, MCARI/MSAVI, CARI/MSAVI) showedthat they improved the predictor accuracy for LAI and were less sensitive to LAI variationunder field condition. The new improve indices proved to be the best predictors of LAI at allgrowth stages with coefficient of determination (R2) ranged from0.683to0.876. The studyresults proved that newly improved hyperspectral spectral indices were better for non-destructive estimation of LAI and monitoring the crop growth in real time. However, thestudy need further testing with the satellite data or airborne imagery for its fully applicationfor non-destructively estimation of LAI.