摘　要：响应“一控两减三基本”的要求，在不减产的前提下减少化肥用量，研究全程机械化大苗机插条件下不同增密减肥处理的水稻叶绿素相对含量（SPAD值）光谱估算模型。分析水稻叶片SPAD值与冠层一阶微分光谱的相关性，建立基于敏感波长及一阶微分光谱参数的SPAD值估算模型，用决定系数（R2）、均方根误差（RMSE）、相对误差（RE）对模型精度进行评价。结果表明：6个处理在634～652 nm、655～668 nm、497～552 nm及687～711 nm波段内达极显著相关，相关系数均达到0.700以上；以敏感波长为自变量估算SPAD值的模型中，低密度减肥的多项式模型估测效果最佳，其建模集R2、RMSE和RE分别为0.766、1.185和2.152%，验证集R2、RMSE和RE分别为0.729、0.364和0.752%；以光谱参数为自变量的估算模型中，中等密度减肥模型的估测效果最好，其建模集R2、RMSE和RE分别为0.703、1.314和2.359%，验证集R2、RMSE和RE分别为0.763、0.620和1.198%。研究发现：在同密度条件下，减肥会导致SPAD值降低，长势偏弱；移栽密度越大，光谱反射率相对越高；移栽密度对水稻冠层光谱反射率的影响大于施肥水平。各处理均表现为差值光谱指数（DSI）的模型效果最优，说明敏感波长及光谱参数的选择对模型的构建至关重要。
关键词：杂交稻；增密减肥；机插；SPAD值；光谱指数
中图分类号：S511；TP7　　　　　　文献标志码：ADOI：10.3969/j.issn.1007-7146.2021.03.007

Abstract: In response to the requirements of “One Regulatory, Two Reduction and Three Basic”, the use of chemical fertilizers was reduced without reducing production. This paper studies the SPAD value spectral estimation model with different increasing density and reducing fertilizer treatments under the condition of full mechanization and machine-transplanted rice with long seeding age and analyzes the correlation between SPAD value of rice leaves and the first derivative spectral reflectance of the canopy spectrum. The SPAD values estimation model were established based on the sensitive wave lengths and the first derivative spectra parameters. The R2, RMSE and RE were used to evaluate the accuracy of the models. The results found that the 6 treatments reached extremely significant correlations in the 634～652 nm, 655～668 nm, 497～552 nm and 687～711 nm bands, all four bands with a correlation coefficient of more than 0.700. Among the models that use the sensitive wavelength as the independent variable to estimate the SPAD value, the polynomial model of low-density fertilizer reduction had the best estimation effect, with 0.766, 1.185 and 2.152% for the calibration set R2, RMSE and RE, and 0.729, 0.364 and 0.752% for the validation set R2, RMSE and RE, respectively. Among the models that use the spectral index as the independent variable to estimate the SPAD value, the medium-density fertilizer reduction has the best estimation effect and its calibration set R2, RMSE and RE were 0.703, 1.314 and 2.359%, respectively, the validation set R2, RMSE and RE were 0.763, 0.620 and 1.198%, respectively. The study found that under the same density conditions, reducing fertilizer would reduce the SPAD value and weaken growth; the higher the transplanting density, the higher the spectral reflectance; the influence of transplanting density on the spectral reflectance of the rice canopy was greater than the fertilization level. All treatments showed that the difference spectral index (DSI) model effect was the best, indicating that the selection of sensitive wavelengths and spectral parameters is critical to the construction of the model.
Key words: hybrid rice; increase density and reduce fertilizer; machine-transplanted; SPAD value; spectral index
（Acta Laser Biology Sinica, 2021, 30(3): 250-258）