Monitoring Wheat Potassium Status Based On Hyperspectral Technology

Quantitative measurement of plant potassium nutrient based on the hyperspectral technology has become an important researching area in remote sensing of vegetation,both in China and abroad.Measuring the potassium nutrient in plants in a fast and non-destructive way is certainly required by developing Precision Agriculture.Studying the wheat as the research subject,we carried out pot cultivation experiments based on varieties,potassium contents and nitrogen levels.By obtaining the hyperspectral information and agricultural data of both single leaf and canopy during critical growth period,we explore the core band and the sensitive parameters indicating the potassium nutrient in wheat.The core band and the parameters are used to help building the wheat potassium nutrient monitoring models applying to different varieties and N levels,which offers technological support to real-time monitoring wheat potassium nutrient in Precision Agriculture.The first step is to find the sensitive band combinations of wheat potassium nutrient in single leaf.In this research,we chose Leaf potassium content(LKC)and K accumulation per unit leaf(which includes the information of leaf area and biomass)as monitoring indices,to reveal the dynamic variation rule of wheat simple leaf potassium nutrient in various potassium and nitrogen levels during the whole growing process.We analyzed the variation characteristics of wheat single leaf spectrum in different potassium and nitrogen nutritional conditions,and analyzed the quantitative relationships of wheat potassium nutrient to NDSI(Normalized difference spectrum index)、RSI(Ration spectrum index)、and DSI(Difference spectrum index)composed of any two wavebands with original reflectance and its derivative within the spectral range of 350-2500nm,and to the three-band spectral index,furthermore we built a relatively universal monitoring model of wheat single leaf potassium.The results show that the wheat leaf spectrum under different potassium conditions varies widely in visible and short-infrared band,which suggests the existence of potassium sensitive wavebands.The potassium sensitive band varies in different nitrogen conditions and we found that N1N2 commonality sensitive band of LKC is λ1=2250-2325nm,λ2=1860-1880nm,and N1N2 commonality sensitive band of K accumulation per unit leaf is λ1＝1285-1 320nm,λ2=1185-1285nm.NDSI(R2275,R1875)、RSI(FD1865、FD2250)and the three-band spectral indices(R2275-R1875)/(R2275+R1875-2R762)、(R2280-R780)/(R1875-R780)can well predict the LKC in wheat,while the original spectral index DSI(R1310,R1210)and the derivative spectral index RSI(FD2250,FD1395)can predict effectively the accumulation of potassium per unit leaf with slight influence of varieties and N factor.The above models are simply sensitive to potassium and have a good stability.The next step is to build the wheat potassium monitoring model using the canopy spectrum.We made an analysis of the wheat canopy spectral variation in different potassium levels,and analyze the relationships between the spectrum and LKC(Leaf potassium content),LKA(Leaf potassium accumulation),K accumulation per unit leaf,ABKC(Aboveground potassium content)as well as ABKA(Aboveground potassium accumulation).The results indicate that the determination coefficients of canopy LKC monitoring models,based on NDSI(R1480,R1435)、DSI(FD1140,FD980),are separately 0.416 and 0.521.The indices NDSI(R975,R905)and NDSI(FD1005,FD690)have a better monitoring precision of LKA and the determination coefficients are 0.707 and 0.779.The indices based on NDSI(R1780,R1630),RSI(FD1545,FD695)have a better monitoring effect on the accumulation of potassium in unit leaf area of wheat.The determination coefficients of ABKC monitoring models,on the basis of the spectral indices NDSI(R935,R770)and DSI(FD1715,FD690),are 0.480 and 0.629.The determination coefficients of the ABKA monitoring models,on the basis of the spectral indices RSI(R610,R510)and DSI(FD630,FD405),are 0.572 and 0.757.Besides,the three-band indices(R975-R905-R1104)/(R974+R905+R1104)、(R610-R510)/(R610+R510-2R435)have an excellent effect on monitoring LKA and ABKA,and compared to the original two-band index,the accuracy has greatly improved.After comparing the precision and the deviation of all models,in the conception of developing a portable potassium monitoring instrument and in principle of remote sensing,we come to the conclusion that it is feasible to use LKA as the wheat potassium monitoring index,and the results may be more accurate and reliable.