Narrow-waveband spectral indices for prediction of yield loss in frost-damaged winter wheat during stem elongation

文献类型： 外文期刊

第一作者： Wu, Yongfeng

作者： Wu, Yongfeng;Ma, Juncheng;Zhao, Haigen;Ma, Ying;Hu, Xin;Ren, Dechao

作者机构：

关键词： Frost treatment; Published indices; Optimized indices; Yield reduction rate; Prediction accuracy; Stem elongation

期刊名称：EUROPEAN JOURNAL OF AGRONOMY （ 影响因子：5.124； 五年影响因子：5.567 ）

ISSN： 1161-0301

年卷期： 2021 年 124 卷

页码：

收录情况： SCI

摘要： Frost during stem elongation is one of the most destructive disasters in China, which has a significant impact on the production of winter wheat. Automatic monitoring of frost injury to canopy is of vital importance to the early prediction of yield loss. This study investigated the potential of hyperspectral techniques in predicting the Percent Yield Difference (PYD) of frost-damaged winter wheat. Three artificial frost experiments were conducted to obtain hyperspectral reflectance and grain yield for winter wheat subjected to sub-freezing temperature treatments. The PYD was used to evaluate the level of frost damage. Nine new indices based on the best combination of wavelengths were selected through the contour mapping approach. All new and published indices were used to establish the linear regression models with PYD. The results showed that as the value of PYD increased, the NIR reflectance within 760-1140 nm decreased, whereas the red and SWIR reflectance increased. The most significant change was found in the NIR region, where the water absorption bands within 930-970 nm almost disappeared. The cross-validation results indicated that the three water-sensitive spectral indices NWI-2, RDSI ((R-958 - R-545)/(R-826 -R-545)), and NDSI ((R-962 - R-829)/(R-962 + R-829)) demonstrated the best prediction accuracy and outperformed the partial least square regression (PLSR) and support vector regression (SVR) models. NWI-2 and NDSI represented a relatively simple waveband combination similar to NDVI, which could be referenced for developing satellite multispectral products to predict PYD at a large spatial scale. GS and PYD range had a significant impact on the spectral indices. The prediction accuracy of PYD for a single GS improved as development advanced before heading. When the PYD value was above 10 %, no significant differences between the subfreezing treatments and the unfrosted controls was detected until the PYD value exceeded 30-40%. It was difficult to predict the relatively low PYD level due to the hybrid response of the spectral reflectance to frost damage.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序