定量描述叶片颜色变化的动态过程是植物生长数字化和可视化的重要内容。本研究通过对不同水稻品种和不同水氮处理条件下主茎和分蘖不同叶位叶色变化过程的连续观测和定量分析，构建了水稻叶片颜色随生长度日变化的动态模拟模型。水稻叶片颜色变化的基本过程可以用3个阶段的SPAD值分别表达，第一阶段为基于幂函数的伸长期，叶色逐渐增强，第二阶段为相对稳定的功能期，叶色基本不变，第三阶段为基于二次曲线的衰老期，叶色逐渐减弱；并基于二次曲线方程分别描述了叶片含氮量和含水量对叶色变化过程的调控效应。在此基础上，进一步建立了叶片SPAD值与叶色组分（RGB）值的关系模型。利用独立的水稻田间试验资料对所建模型进行了测试和检验，显示主茎不同叶位叶色变化3个阶段模拟值的均方根差分别为2.58、3.69和3.82，4个分蘖不同叶位叶色变化模拟值的均方根差分别为4.65、4.39、3.51和4.25；SPAD值与叶色组分间模拟值的均方根差分别为2.98和3.25。表明本模型可较好地模拟不同生长条件下水稻不同茎蘖上不同叶位叶色的动态变化过程，从而为实现水稻生长系统的数字化模拟和可视化显示奠定了基础。
With the development of digital agriculture, the visualization expression of plant growth process is very important. Quantifying the dynamic process of leaf color is the basic part of plant growth visualization. At present, the study on plant leaf color visualization at home and abroad is emphasized on reversing and rebuilding crop virtual growth with the figure identifying technique in computer, however, simulation model and visualization technique based on leaf color dynamic process have not been established successfully.
Modeling leaf color dynamics in rice is an important task for realizing virtual and digital plant growth. Based on time-course observations on leaf color changes at different leaf positions of stem and tillers under different nitrogen rates and water conditions with four rice cultivars, we developed a simulation model on leaf color dynamic changes in rice in relation to GDD. Leaf color changes in the model were described with SPAD in three phases. The first phase during leaf growth period was based on the exponential relationship of leaf color to cumulative GDD; the second phase during leaf function period was represented with a relative stable SPAD; the third phase during leaf senescence period was described in a quadratic equation between SPAD and GDD. In addition, the effects of nitrogen and water conditions on leaf color were quantified through the effectiveness values of leaf nitrogen concentration and water content in relation to SPAD. Then, the RGB values were further predicted from the changing SPAD. The model was validated with the independent field experiment data involving different rice cultivars and nitrogen rates. The average RMSEs between the simulated and observed SPAD dynamics at different leaf positions were 2.58, 3.69, and 3.82, respectively, for three leaf color phases on main stem, 4.65, 4.39, 3.51, and 4.25, respectively, for four individual tillers in rice, and 2.98, 3.25, respectively, for SPAD and R, G values.
The results indicate that the present model has a good performance in predicting leaf color changes at different leaf positions in rice under different growth conditions, and thus lays a foundation for further constructing digital and visual rice growth system.
全 文 ：