Monitoring Wheat Leaf Nitrogen Nutrition With Sun-Induced Chlorophyll Fluorescence

With the advancement of agricultural and rural pollution prevention and control battles,in order to reduce soil and water pollution,it is essential to controll nitrogen fertilizer input and manage field nitrogen fertilizer precisely in wheat field.Monitoring the wheat nitrogen nutrition accurately and non-destructively is of great significance to guide field nitrogen fertilizer management,guarantee crop yield and quality,reduce environmental pollution and improve economic benefits.Traditional optical remote sensing technology has been widely used to monitor the nitrogen nutrition of wheat.But the reflectance signal is extracted from the mixed pixels of vegetation and soil background,which is easily affected by the soil background and is not sensitive to nitrogen stress.While sun-induced chlorophyll fluorescence(SIF)closely related to photosynthesis,can respond to environmental changes quickly and sensitively,and provide reliable technical support for monitoring nitrogen nutrition,estimating crop yield and diagnosing stress status.In this study,four wheat field experiments were conducted,including different growing seasons,wheat cultivars,planting densities and nitrogen treatments.The SIF signal,reflectance,nitrogen indices,photosynthesis and chlorophyll parameters were collected at the leaf and canopy scales to achieve the following three research goals.The research results would provide theoretical and technical support for monitoring and diagnosing wheat nitrogen nutrition using SIF.At the leaf and canopy scales,the leaf nitrogen content index(LNC)was selected by clarifying the seasonal variation of area/mass based LNC and SIF signal,and analyzing the quantitative relationships among them.The most optimal estimation model based on SIF indices for monitoring LNC was constructed.The results showed that at the leaf and canopy scales the area-based LNC was more related to SIF signal than mass-based LNC by comparing the seasonal variation and estimation accuracy,especially at the canopy scale.The accuracy of the SIF indices,SIF ratio index(↑SIFR2)and the normalized SIF index(↑SIFN),constructed by two SIF bands at 687 and 761 nm for estimating area-based LNC were higher than the single-band SIF index and vegetation index(leaf scale:calibration R2(0.75,0.74),validation R2(0.66,0.55);canopy scale:calibration R2(0.87,0.86),validation R2(0.65,0.62)).At the same time,the performance of ↑SIFR2 and ↑SIFN to monitor photosynthesis nitrogen use efficiency(PNUE)were also significantly better than vegetation indices.In addition,by the analysis in the entire range and five fixed ranges of data,it was found that the reason why ↑SIFR2 and ↑SIFN can monitor the area-based LNC accurately is not only plant traits(leaf scale:area-based leaf chlorophyll content(LCC),leaf mass per area(LMA);canopy scale:canopy area-based LCC,leaf area index,leaf dry weight per unit soil area,and LMA),but also other potential internal factors.At the leaf scale,the characteristics change of the upward and downward SIF yield spectrum with the area-based LNC(LNCarea)were explored,the differences between the upward and downward SIF yield index for monitoring LNCarea and the reasons were compared,and the best estimation model for LNCarea were established(it was comfirmed that LNCarea was more related to SIF signal in the former chapter,so only the LNCarea was considered in this and later chapters).The results showed that the signal of upward SIF was stronger than that of downward SIF,due to the chlorophyll reabsorption and scattering effects inside the leaf.The SIF peak in the red region was smaller than that in the far-red region.Compared with the upward SIF yield indices,the accuracy of the downward SIF yield indices for monitoring LNCarea were higher.The performance of SIF peak ratio index based on two SIF peaks in red and far-red region(↓ SIFR1 and ↓SIFR1)for detecting LNCarea were better than that on single SIF peak(calibration R2 were 0.52 and 0.71;validation R2 were 0.51 and 0.74,validation RRMSE were 23.09%and 16.95%,respectively),especially ↓SIFR1.In addition,the LNCarea models based on ↑SIFR1 and ↓SIFR1 performed stably,which were not affected by LCC and LMA.At the leaf and canopy scales,the response regular of time series SIF indices,vegetation indices,agronomic parameters and leaf absorption light energy distribution parameters with wheat nitrogen stress were compared.The response mechanism of various parameters was analyzed.The SIF index most sensitive to nitrogen stress was selected,on which a wheat NNI monitoring model were constructed.The results showed that the SIF index increased firstly and then decreased with the nitrogen stress,and reached the maximum around day after treatment(DAT)12.↑SIFR2 responded to nitrogen stress on DAT 4 at the leaf and canopy scales,which is the most timely.The red edge chlorophyll index(CIred edge)was the most sensitive vegetation index to nitrogen stress,and the response times are DAT 12 and DAT 8 at two observational scales,respectively.Among the agronomic parameters,the most sensitive index to nitrogen treatments was the leaf photosynthesis,which responses on DAT 4.The leaf dry matter(LDM),plant dry matter(PDM)and LAI under different nitrogen levels showed significant differences after DAT 8.The area-based LCC responded to nitrogen stress after DAT 12.Meanwhile,the temporal change of the parameters related to the leaf absorbed light energy distribution were shown as follows:in the normal nitrogen application range,the quantum yield of photochemistry increased with the increase of nitrogen application,and the sum of the quantum yield of fluorescence and constitutive heat dissipation decreased.With excessive nitrogen application,the quantum yield of photochemistry didn’t increase significantly.The fluctuation of the quantum yield of nonphotochemical quenching in each nitrogen treatment wasn’t large.Moreover,by the temporal change of SIF indices,agronomy parameters,and vegetation indices under different nitrogen levels,it was determinated that ↑SIFR2 responded to the nitrogen stress most timely.And based on the most sensitive SIF index ↑SIFR2,the NNI model based on LDM and PDM were establied with the coefficient of determination of 0.71 and 0.63,and the root mean square error of 0.08 and 0.06,respectively.