Abstract Monitoring the chlorophyll content in tuber formation and tuber bulking stage has great significance for the nutritional status diagnosis in the potato field. In this paper, the 314 canopy spectra of potato crops were collected at four stages, respectively. The leaves were collected simultaneously to measure the chlorophyll content. After spectra pre-processing, the dynamic changes in chlorophyll content and spectral response during growth were analyzed. Two variable selection algorithms (successive projection algorithm, and random frog algorithm) were employed to select chlorophyll characteristic wavelengths. The partial least square algorithm was used for modeling analysis. The performance of sensitive wavelengths selected by two algorithms was compared, respectively, from the perspective of correlation with chlorophyll, reflecting leaf information, and model result. The sensitive wavelengths selected by random frog was optimal. The determination coefficient of calibration and validation set of the detection model established based on above sensitive wavelengths was respectively 0.827 and 0.798. The results showed the chlorophyll content of potato could be detected accurately based on the spectroscopy method.

中文翻译：

---

基于光谱变量分析的生长马铃薯叶绿素含量检测

摘要 块茎形成和块茎膨大期叶绿素含量监测对马铃薯田间营养状况诊断具有重要意义。本文分别在四个阶段收集了马铃薯作物的314个冠层光谱。同时收集叶子以测量叶绿素含量。光谱预处理后，分析了生长过程中叶绿素含量的动态变化和光谱响应。采用两种变量选择算法（连续投影算法和随机青蛙算法）来选择叶绿素特征波长。采用偏最小二乘算法进行建模分析。分别从与叶绿素相关性的角度比较了两种算法选择的敏感波长的性能，反映叶子信息，和模型结果。随机青蛙选择的敏感波长是最佳的。基于上述敏感波长建立的检测模型的校准和验证集决定系数分别为0.827和0.798。结果表明，基于光谱法可以准确检测马铃薯叶绿素含量。