目录
摘要Ⅱ
关键词Ⅱ
AbstractⅢ
引言
引言1
第一章 材料与方法4
1.1试验设计 4
1.2测定项目与方法4
1.2.1冠层高光谱反射率测定4
1.2.2农学参数测定4
1.3模型构建及精度检验5
第二章 结果与分析6
2.1不同氮肥处理下的水稻冠层高光谱反射率特征...6
2.2水稻产量及地上部氮积累量的高光谱预测...6
2.2.1水稻产量和地上部氮积累量与植被指数的相关性6
2.2.2水稻产量和地上部氮积累量估算模型及精度检验7
2.2.2.1水稻产量常规回归模型及精度检验7
2.2.2.2基于随机森林算法的水稻产量估算模型及精度检验8
2.2.2.3地上部氮积累量常规回归模型及精度检验10
2.3水稻氮素籽粒生产效率的间接和直接高光谱预测11
2.3.1水稻氮素籽粒生产效率的间接高光谱预测11
2.3.2水稻氮素籽粒生产效率的直接高光谱预测12
第三章 讨论14
致谢15
参考文献15
水稻氮素籽粒生产效率的高光谱遥感反演
摘要
高光谱分析技术凭借其快速、无损的监测优点，已被应用于水稻产量及生理生化参数的反演中。为快速、准确地实现对水稻氮高效品种的筛选，开展了基于高光谱成像技术的常规水稻的氮素籽粒生产效率的间接和直接估算研究。本文利用两年试验数据建立了基于SAVI（R730, R982）的氮素籽粒生产效率直接估算模型和基于产量和植株地上部氮积累量预测值的水稻氮素籽粒生产效率间接估算模型。结果表明：（1）通过随机森林算法建立的以分蘖期DVI（R540, R697）、抽穗期SAVI（R731, R730）、成熟期DVI（R845, R760）为自变量的多时相估算模型（R2=0.86，RMSE=49.81，RE=11.43%）较单生育期估算模型的预测精度更高;（2）以SAVI（R730, R982）为自变量构建的二次多项式模型是直接预测成熟期植株地上部 *51今日免费论文网|www.jxszl.com +Q: #351916072# 
氮积累的最佳单变量模型（R2=0.90，RMSE=2.94，RE=27.24%）;（3）氮素籽粒生产效率的直接估算模型（R2=0.75，RMSE=10.36，RE=17.01%）的验证R2和RE相较于间接估算模型(R2 = 0.63，RMSE=19.6，RE=13.03%）略高，RMSE略低，两者预测精度相差不大，因此两者都可用于植株氮素籽粒生产效率的估算和氮高效品种的筛选中。
HYPERSPECTRAL ESTIMATION MODEL FOR NITROGEN PADDY PRODUCTION EFFICIENCY
ABSTRACT
Hyperspectral analysis technology has been applied to the monitoring of rice physiological and biochemical parameters with its advantages of fast and nondestructive monitoring. The indirect and direct estimation of rice nitrogen paddy production efficiency which is based on the hyperspectral imaging technology was launched on the purpose of high throughput screening of nitrogen efficient rice varieties. In this paper, two years of experimental data was used to establish the direct model of nitrogen paddy production efficiency based on SAVI (R730, R982) and indirect model based on the estimation model of yield and plant nitrogen accumulation. The result shows that: (1) the prediction accuracy of random forest model (R2=0.86, RMSE=49.81, RE=11.43%) using multitemporal vegetation indices (DVI (R540, R697) , SAVI (R731, R730) , DVI (R845, R760) at tillering, booting, heading stage, respectively) is higher compared to estimation model of single stage; (2) quadratic polynomial model with SAVI (R730, R982) as independent variable performs best for the accumulation of plant nitrogen at maturation stage (R2=0.90, RMSE=2.94, RE=27.24%) ; (3) the direct estimation (R2=0.75, RMSE=10.36, RE=17.01%) of nitrogen paddy production efficiency has a slight lower R2 and RE than the indirect model (R2 = 0.63, RMSE=19.6, RE=13.03%) , but with higher RMSE. There is little difference in prediction accuracy, so both of them can be used to predict nitrogen paddy production efficiency and screen nitrogen efficient varieties.

原文链接：http://www.jxszl.com/nongxue/zwbh/606193.html