Based On Hyperspectral Remote Sensing Of The Rice Growth Monitoring Research

Fast, real-time, accurate access to farmland ecological environment and crop growth information is an important basis premise for the implementation of precision agriculture, it is also one of the key technology bottleneck of modern precision agriculture development. In this paper, with rice as the research object, based on the community experiments of different fertilization(nitrogen fertilizer, biomass carbon fertilizer) gradient, the integrated use of high spectral resolution remote sensing, physiological and biochemical parameters testing and mathematical statistics and other technical means, analysis under the different fertilization(nitrogen fertilizer, biomass carbon fertilizer) the rice crown height spectrum characteristics, chlorophyll content characteristics and leaf area index characteristics of different development stages. Then the hyperspectral estimation models of chlorophyll content, LAI established on the basis of original spectral reflectance parameters, "Trilateral" parameter at different growth stages. And use the determination coefficient(R2), root mean square error and relative error evaluation index to validate the precision of the prediction model. This research mainly get the following results:(1) Changes of canopy spectral in different growth period and different fertilization(nitrogen fertilizer, biomass carbon fertilizer) were studied. The result showed that from jointing stage to milk stage, rice canopy spectral reflectance in the visible area is increasing, and in the near infrared region is first increases then decreases. With the increase of nitrogen level, rice canopy spectral reflectance in the visible light range is overall present the trend of gradually reduce, however rice canopy spectral reflectance in near infrared band is increased gradually. Different level of the biomass carbon, the difference of rice canopy spectral reflectance is not obvious in visible band, however arrived at near infrared wavelength, the difference of canopy spectral reflectance is obvious and the rice canopy spectral reflectance values of carbon processing greater than canopy spectral reflectance values without carbon trading. Therefore, the research results provide certain theoretical basis for monitoring the rice growth conditions of deeper using rice canopy spectral information.(2) The correlationship between rice canopy original spectrum, derivative spectrum and rice chlorophyll content, LAI were analyzed respectively. In visible light region, the rice original spectral reflectance of jointing stage, heading stage and filling stage was negatively correlated with chlorophyll content, LAI. In the "red edge", the correlation from negative to positive. The correlation coefficients of derivative spectrum and chlorophyll content, LAI were higher than those of original spectrum and chlorophyll content, LAI at some wavelengths in jointing stage, heading stage and filling stage. There were low correlationship between original spectrum, derivative spectrum and chlorophyll content, LAI at ratooning buds. Therefore, when established the high spectral estimation model of rice chlorophyll content, LAI, the spectral data in the ratooning buds could not be used.(3) The data including rice canopy spectrum, chlorophyll content, leaf area index in 2014, were used to establish the estimation models based on original spectral reflectance parameters(green peak position, green peak reflectance, green peak area, red valley position, red valley reflectance, red valley area, the ratio and normalization value of green peak area and red valley area), "Trilateral" parameter(blue edge position, blue edge amplitude, blue edge area, yellow edge position, yellow edge amplitude, yellow edge area, red edge position, red edge amplitude, red edge area, the ratio and normalization value of red edge area and blue edge area, the ratio and normalization value of red edge area and yellow edge area) of rice chlorophyll content, LAI. The precision of established models was tested by using the data of rice canopy spectrum, chlorophyll content and LAI in 2013. The results showed that when using the original spectral reflectance parameters inversion chlorophyll content of rice, the red valley area should have the top priority in jointing stage. The inversion effect of red valley reflectance, red valley area was very good in heading stage and filling stage. When retrieving LAI inverted by the original spectral reflectance parameters, the normalization value of green peak area and red valley area should be the prior considerations in the jointing stage. The inversion effect of red valley reflectance, red valley area was relatively good in heading stage. The inversion effect of green peak reflectance, red valley area was relatively good in filling stage. When inverting the chlorophyll content based on the "Trilateral" parameter, the ratio value of red edge area and blue edge area should be taken into account first in jointing stage. In heading stage, should give priority to the blue edge area. The inversion effect of the normalization value of red edge area and blue edge area was better. For the inversion of LAI, the inversion effect of red edge area, the normalization value of red edge area and blue edge area was very good in the jointing stage, red edge area should be a prime considerations in heading stage, but in filling stage, the inversion effect of red edge position, the ratio value of red edge area and blue edge area was relatively good.