摘要
为实现高光谱技术快速检测空心菜镉污染,探索性地对其叶绿素质量分数的变化情况进行了高光谱估算.采用盆栽试验检测了空心菜在6个不同的镉质量分数梯度0mg/kg(CK),0.5mg/kg(T1),1mg/kg(T2),5mg/kg(T3),10mg/kg(T4),20mg/kg(T5)下的叶片高光谱反射率及其叶绿素质量分数.根据不同镉质量分数处理下空心菜叶片高光谱反射率的变化,对光谱数据重采样,分析了叶绿素质量分数与光谱数据的相关性,筛选出不同质量分数镉处理情况下与叶绿素质量分数敏感的光谱波段及特征参数,并建立了预测叶绿素质量分数的估算模型.研究结果表明:1)随着镉质量分数的增加以及胁迫时间的延长,总叶绿素质量分数呈先增加后减少的趋势.15d,30d和45d的叶片光谱变化趋势相似,在绿峰和红外波段范围差异较明显.处理45d时,其差异比15d和30d时更显著;2)重采样后绿峰和红边范围的波段与叶绿素质量分数具有较好的相关性(p<0.01),相关性最优的波段R540,R710为自变量与叶绿素质量分数拟合的R2值分别为0.747 7,0.713 5(p<0.01),故选择其作为敏感波段;3)对12个光谱敏感参数、R540和R710与叶绿素质量分数进行相关分析、建模并反演,得出R540,R710,GPR和SGP对应的非线性倒数模型效果较优,可用于空心菜叶绿素质量分数的估算.研究结果可进一步为检测空心菜镉污染程度提供技术参考.
In order to rapidly and non‐destructively monitor heavy metal contamination of the leafy vegeta‐ble L pomoea aquatica F. ,a potted experiment was conducted .Six different Cd concentrations treatments (CK ,T1 ,T2 ,T3 ,T4 and T5) were established with 3 replicates per treatment ,where CK ,T1 ,T2 ,T3 , T4 and T5 represented 0 ,0.5 ,1 ,5 ,10 and 20 Cd mg/kg dry soil .The hyperspectral reflective rate of fif‐ty‐four samples was scanned by ASD FieldSpec 3 HR and their chlorophyll concentrations were measured using Shimadzu UV2550 spectrophotometer on the15th ,30th and 45th days after the beginning of treat‐ments .Further ,the correlation between spectral data and chlorophyll contents was analyzed ,and the fit‐ting models for chlorophyll contents were established using the sensitive spectrum and characteristic pa‐rameters .The results showed that total chlorophyll content decreased gradually with higher Cd concentra‐tion and longer treatment time .The differences of leaf reflectance among different treatments were signifi‐cant at green peak and infrared wavelength .In the wave range of 350 -1 300 nm ,the leaf reflectance showed a similar trend between the 45d and other treatments .However ,the difference among the plant leaf reflectance under different Cd concentrations was more significant in the 45‐d treatment ,as compared to that in 15‐d and 30‐d treatments .In peak green and red edge range of wavelengths ,the hyperspectral re‐flectance was well correlated with chlorophyll contents (p<0.01) ,and R2 of R540 and R710 with chloro‐phyll content was 0.753 and 0.719 respectively (p <0.01) after data re‐sampling .Therefore ,R540 and R710 were extracted as the sensitive bands to construct estimation models .The linear ,logarithmic ,in‐verse and parabola fitting models between R540 ,R710 ,G PR and SG P and chlorophyll contents had bet‐ter sensitivity compared with other spectral parameters ,and the calibration R2 and verification R2 were ap‐proximately 0.75 and 0.7 ,respectively .In particu
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2015年第4期9-16,共8页
Journal of Southwest University(Natural Science Edition)
基金
重庆市自然科学基金(cstc2012jjA80003)
重庆市基础与前沿研究计划重点项目(CSTC2013JJB00004)
中央高校基本科研业务费专项资金(XDJK2013A011)
关键词
空心菜
高光谱反射率
敏感参数
估算模型
Lpomoea aquatica F.
hyperspectral reflectance
sensitive parameter
estimation model
