Abstract:

The ability to non-destructively monitor crop water status by remote sensing is of significant importance for optimizing of crop irrigation systems and precision farming. In order to develop quantitative models for monitoring rice water status by canopy reflectance spectra, an investigation was conducted to study the relationship between canopy reflectance characteristics and plant water status under different water and nitrogen levels with different pot and pool experiments. The results showed that the crop spectral reflectance decreased in the visible wavelength regions and short-wave infrared (SWIR) (1 400-1 700 nm) regions with increasing water supply, because absorption of visible light (450-700 nm) by leaf chlorophyll was aggravated whereas the spectral reflectance in the near infrared (NIR) (750-950 nm) wavelength regions increased. A high reflectance spectra flat roof formed which remained throughout the entire growth period. Under different water regimes, canopy reflectance showed similar patterns of change. Under the same water supply, the canopy spectral reflectance in the visible wavelength and SWIR regions was lower under high nitrogen supply than under low nitrogen supply rates but showed an opposite pattern in the NIR wavelength regions. Canopy reflectance in single bands is often subject to interference by biomass, background and so on, but the ratio of two bands can alleviate interference due to topography and enlarge the difference of spectral reflectance between vegetation types thus improving the precision of estimates by spectral reflectance. Consequently, in order to eliminate some of the interferences for distinguishing leaf and canopy structure and establish the best predictive model for assessing leaf water content, all regressions between ratio indices, normalized difference indices and leaf and plant water content were conducted. The results indicated that there was a linear relationship between the vegetation index ratio (R 810 /R 460 ) and canopy leaf water content and plant water content after jointing stage regardless of leaf nitrogen levels, with an RMSE of 0.93 and 1.50, respectively, for model testing. It is concluded that the vegetation index ratio of R 810 /R 460 can be used to monitor leaf water content and plant water content at different growth stages of rice.

Key words: Rice, Soil water, Plant water content, Spectral reflectance, Ratio vegetation index, Monitoring model