Study On Hyperspectral Estimation Of Physiological Parameters Of Rice Under The Damage Of Rice Leaf Roller

Rice is the main food product in China and its importance to China’s agricultural production is self-evident.Cnaphalocrocis medinalis（Rice Leaf Folder）is a common pest that has been troubling rice cultivation in China.In this experiment,the second-instar Rice Leaf Folder and Nanjing 5055 were selected as the research objects.In the agricultural experiment field of Nanjing University of Information Science and Technology,five insect pest classes were set by net control,and the hyperspectral parameters and physiological parameters of rice were measured.The original canopy spectra,first derivative spectra,vegetation index and physiological parameters of rice were analyzed,and the inversion estimation model of relative chlorophyll content and leaf rolling rate was established.The following main research results were obtained:（1）With the improvement of the treatment level,the spectral reflectance and the first derivative of rice canopy decreased gradually（except for the 4th grade treatment）,and the 4th grade treatment was lower than the 5th grade treatment.The 4th grade treatment was the minimum value in this experiment.As the growth period progressed,spectral reflectance increased from tillering stage to booting stage,and then decreased to maturity stage.The reflectance at maturity stage was much lower than that at other growth stages.With the increase of the treatment grade,the vegetation index decreased continuously in the tillering stage,jointing stage,filling stage and maturity stage,while the change range of the vegetation index was smaller during the booting stage and flowering stage.Comparing the treatments of various levels,the vegetation index during the whole growth period of rice was in an inverted parabola shape.Vegetation index was the largest at the booting stage and the smallest at the maturity stage.（2）With the development of growth stage,the net photosynthetic rate（Pn）,stomatal conductance（GS）,intercellular CO₂ concentration（Ci）,transpiration rate（Tr）,relative chlorophyll content（SPAD）and leaf area index（LAI）of rice leaves all increased at first and then decreased,and the transition growth stage was booting stage or flowering stage.The leaf rolling rate kept going up.Physiological parameters changed with the deepening of insect pest level,and the maximum variation range was found in booting stage and 4th grade treatment.（3）After comprehensive analysis of the spectrum and physiological parameters,it was found that the higher pest level didn’t mean the greater damage to rice.The damage of Rice Leaf Folder was the most serious under the 4th grade treatment,and the 4th grade treatment might be the extreme level.During the whole growth period,the parameters often changed in the opposite way before and after booting stage,and booting stage was the key period of this experiment.（4）For SPAD,the model established based on the spectral characteristic bands at the jointing stage and booting stage had the best estimation effect,and the determination coefficient R² reached 0.430 and 0.468.The universality of the period model was better.The other four growth periods were based on the non-linear model of BP neural network,and the prediction effect was better.R² reached above 0.681.For leaf rolling rate,a non-linear model based on BP neural network was selected,and the R² of the model for all growth periods reached above 0.782,and the prediction effect of the model was better.