| 摘要: |
| 【目的】研究西北地区不同生育时期油菜叶片SPAD值与多种光谱参数之间的关系，并建立估测模型。【方法】以陕西省乾县齐南村施肥程度一致的大田种植的油菜为试验材料，分别在苗期、现蕾抽苔期和开花期测定叶片的光谱反射率和SPAD值，得到不同波段组合的光谱指数，计算其与SPAD值之间的相关性系数，并拟合以各种光谱指数预测SPAD值的模型，以均方根误差(RMSE)和相对误差(RE)作为模型的评价和检验指标。【结果】苗期SDr/SDy与叶片SPAD值相关性最高，为0.79，构造的最佳预测模型决定系数为0.67；现蕾抽苔期(SDr－SDy)/(SDr＋SDy)与叶片SPAD相关性最高，达到0.82，构造的最佳SPAD估测模型决定系数为0.70；苗期(SDr－SDy)/(SDr＋SDy)与叶片SPAD有较高相关性，相关性系数为0.65，构造的预测模型决定系数为0.43。【结论】各时期预测模型效果较好的光谱指数基本一致，为SDr/SDy和(SDr－SDy)/(SDr＋SDy)，可以利用高光谱遥感信息反演西北地区油菜各时期的叶绿素含量，并分析其生长状况。 |
| 关键词: 高光谱指数 油菜 SPAD值 预测模型 |
| DOI： |
| 分类号: |
| 基金项目:国家“863”高技术研究发展计划项目（2013AA102401-2） |
|
| Estimation of rape leaf SPAD in different periods based on spectral indices |
|
YIN Zi,CHANG Qingrui,LIU Miao,et al
|
| Abstract: |
| 【Objective】This study analyzed the correlations between SPAD values and various spectral parameters of rape leaf in different stages in northwest China,and established models for estimating SPAD. 【Method】This experiment was conducted in Qinan village,Qianxian,Shaanxi.Rape was planted in fields with same fertilization level.Spectral reflectance and SPAD values of leaf in different periods were measured and hyperspectral indices were obtained.Then correlation coefficients between them were calculated and prediction models were established.The accuracy of models was validated using root mean square error (RMSE) and relative error(RE).【Result】SDr/SDy had the highest correlation level (R=0.79) with SPAD in seedling stage,and the R2 of the best estimation model was 0.67.(SDr－SDy )/(SDr＋SDy) had the highest correlation level (R=0.82) with SPAD in budding and bolting stages,and the R2 of the inversion model was 0.70.(SDr－SDy )/(SDr＋SDy) had the highest correlation level (R=0.65) with SPAD in anthesis stage,and the R2 of the inversion model was 0.43.【Conclusion】The models established with SDr/SDy and (SDr－SDy)/(SDr＋SDy) had the best results.The results could provide basis for estimating rape chlorophyll content in different periods and analyzing its growth in northwest China. |
| Key words: hyperspectral parameters rape SPAD estimation model |
