叶龄模型在水稻上应用的检验与比较

Validation and comparison of the precision of leaf emergence models on rice

为筛选出适用于水稻出叶动态模拟的叶龄模型,利用江苏南京和云南丽江2个生态点、多个品种、多个播期的水稻出叶动态数据和温光资料,对前人提出的基于热时间的线性模型、基于热时间的幂函数模型、基于辐热指数的线性模型、水稻钟模型和Beta模型等5个模型对水稻出叶动态的描述能力和预测能力进行了检验与比较。结果表明,Beta模型、水稻钟模型和基于热时间的幂函数模型对不同品种在不同生态条件下出叶动态的拟合具有较高的精度,基于热时间的线性模型和基于辐热指数的线性模型拟合精度较低;3个非线性模型中,基于热时间的幂函数模型对不同生态点水稻出叶动态均有较好的预测效果,而水稻钟模型和Beta模型能较准确地模拟一般生态条件下水稻的出叶动态,但对于丽江试验点这一特殊生态点水稻出叶动态的模拟效果欠佳。由此可见,水稻出叶速率在整个生育进程中不是恒定的,对水稻出叶动态进行模拟不宜选用基于热时间的线性模型和基于辐热指数的线性模型等线性模型,而应选用出叶速率随生育进程逐渐降低的基于热时间的幂函数模型、水稻钟模型和Beta模型等非线性模型;由于特殊的温光条件使品种的出叶进程对时间的响应发生了较大的改变,使得以出苗后时间为驱动变量的水稻钟模型和Beta模型准确度下降;而基于热时间的幂函数模型在描述能力、预测准确性、适用范围等方面均表现较好。

To select leaf age models suitable for rice, description ability and prediction ability of previous proposed thermal time based linear model （TTLM）, thermal time based power model （TTPM）, photo-thermal index based linear model（PTILM）, rice clock model （RCM）, and beta function based model （BBM） were tested and compared using data of leaf emergence, air temperature and light condition from experiments differed in sowing dates and varieties at two ecological sites, Nanjing Jiangsu province, and Lijiang, Yunnan province. The BBM, RCM and TTPM fitted rice leaf emerging dynamics with higher precision than TTLM and PTILM. Of the three nonlinear models, TTPM gave good results while used forpredicting rice leaf emerging dynamics at all experimental sites, the RCM and BBM accurately simulated the dynamics of rice leaf emergence only at Nanjing. It can be concluded that nonlinear models of TTPM, RCM and BBM are more suitable for the description of the dynamics of rice leaf emergence than linear models of TTLM and PTILM due to inconstant leaf emergence speed. In addition, since the special temperature and light conditions in Lijiang experi-mental site may change the response of leaf emergence process to days after emergence（DAE）, the accuracy of BBM and RCM which using DAE as driving variable dropped; However, the TTPM based on thermal time performs better in description ability, predicting accuracy and applicable scope.