基于高斯函数的水稻叶曲线动态模拟

【目的】以高斯函数为基础构建空间曲线模型，模拟研究水稻叶曲线空间变化特征。【方法】基于不同年份与施氮水平的水稻桶栽试验，定期利用三维激光扫描仪FastScan获取水稻主茎不同叶位叶曲线空间坐标数据，并利用动态建模技术构建水稻主茎不同叶位叶曲线动态模拟模型。【结果】水稻主茎不同叶位叶曲线末端与y轴正方向的夹角（AH）随生长度日（GDD）呈快-慢-快的变化趋势，符合S型曲线；从第1叶到第7叶，AH的最大值随叶位增加而增加，随后随叶位增加而减小；鞘叶夹角∠BFC与AH呈极显著线性关系。本文利用Logistic方程分别描述了不同施氮水平下水稻主茎各叶位叶曲线对应的AH以及叶曲线模型参数Sm随生长度日GDD的变化；使用分段函数描述了AH最大值随叶位的变化；将适宜施氮条件下水稻主茎第7叶位叶曲线末端与y轴正方向的夹角作为品种参数量化品种对叶曲线的影响，引入氮素影响因子量化不同氮素水平对叶片披垂度的影响。利用独立田间试验资料对所建立的模型进行检验后显示，不同施氮水平下主茎不同叶位叶曲线实测点到模拟曲线的改进Hausdorff距离（LTS-HD）的均值在分蘖中期和拔节期分别低于0.88 cm和1.18 cm。【结论】模型对水稻主茎叶曲线的空间动态变化过程具有较好的预测性，研究结果为进一步提高水稻叶片及植株的可视化效果提供了技术支持。

Dynamic Simulation on Leaf Curve in Rice Based on Gaussian Function

【Objective】 The dynamic changes in leaf curve of rice under different nitrogen levels were analyzed using Gaussian function. 【Method】Based on rice pot experiments involving different years and nitrogen levels, a 3-D laser scanner was used to measure the spatial coordinate data of leaf curve at different leaf positions on main culm of rice under different nitrogen levels, then dynamic modeling technology was used to build the model of leaf curve changes at different leaf positions on main culm in rice. 【Result】 The results indicated that the angle between the end of leaf curve and positive direction of y-axis (AH) increased with growth degree days on the trend of slow-fast-slow, which could be described with the Logistic function. From the first leaf to the seventh leaf, the maximum value of AH increased with increasing leaf position, and then decreased with increasing leaf position. The significant linear relationship could be observed between sheath-leaf angle∠BFC and AH. The logistic equations were used to describe the changes of AH and the model parameter Sm in leaf curve model with growth degree day (GDD), respectively. The piecewise function was applied to describe the change of maximum AH with leaf position. A variety parameter (AH of the seventh leaf position on main culm in rice under optimal nitrogen conditions) and nitrogen factor were introduced to quantify the effects of variety and nitrogen rate on leaf drooping degree, respectively. Model validation with independent field experiment data showed that the average value of LTS-HD (least trimmed squares Hausdorff distances) between observed and predicted curve are less than 0.88 cm and 1.18 cm at tilling and jointing stages, respectively. 【Conclusion】 The model showed a good predictability for spatial dynamic change of leaf curve on main culm under different nitrogen rates during main growth stages in rice. These results would provide a technical support for visualization of leaf and plant in rice.