期刊
FIELD CROPS RESEARCH
卷 214, 期 -, 页码 73-82出版社
ELSEVIER
DOI: 10.1016/j.fcr.2017.08.023
关键词
Fertilization management; Soil N; Multi-model selection; NNI; SPAD
类别
资金
- Arvalis-Institut-du-vegetal
- ADEME (Agence de l'Environment et de la Maitrise de l'Energie)
We investigated the feasibility of predicting crop nitrogen nutrition index (NNI) from chlorophyll meter readings (CMRs) during the crop cycle, for the development of a fertilization method for winter wheat (Triticwn aestivwn L.) based on the regular monitoring of crop N status. The relationship between NNI and CMR has been studied before, but only for CMRs obtained late in the season. A literature review revealed an absence of consensus concerning the most accurate equation for predicting NNI from CMR. It remains unclear which variables are the most influential and the extent to which it might be possible to overcome these uncertainties by using a normalized chlorophyll meter reading. We therefore carried out multimodel selection, comparing goodness-of-fit, prediction accuracy and likelihood, for linear, quadratic and exponential models, taking into account only biomass, cultivar, biomass plus cultivar and growth stage effects. Models were fitted with absolute and normalized measurements. We also considered the possibility of predicting NNI with a single model or with different models for each growth stage. We found that normalized measurements limited the biomass and cultivar effects, but not the growth stage effect. Furthermore, the use of normalized measurements increased prediction accuracy. However, the prediction error remained very high if the well-fertilized strip was N-deficient (NNI < 0.9). Finally, the best compromise was found to be a model for each growth stage, using absolute measurements, but taking into account the effects of biomass and cultivar.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据