水浇地与旱地春小麦新型光谱曲线特征变量与叶绿素关系的对比研究

为了利用新型光谱曲线特征变量对水浇地与旱地春小麦叶绿素含量进行估算，测定水浇地与旱地春小麦冠层光谱、叶绿素含量、叶宽和苗高，并利用各地春小麦冠层光谱曲线“峰谷”存在差异的特点，通过分析光谱曲线特征变量与水浇地和旱地春小麦叶绿素含量的相关性，分别建立了水浇地和旱地春小麦叶绿素含量最佳估算模型并检验模型精度。结果表明：在723 nm、730 nm和736 nm前后，水浇地、阴坡地、阳坡和双面坡地春小麦原始光谱与叶绿素含量的相关性表现出完全相反的变化规律。在以上波段之前，相关系数为负，之后为正。各地春小麦叶绿素含量与绿边和黄边曲线变化速率及反射绿峰两边的夹角存在很好的相关性，与水浇地、阴坡地、阳坡和双面坡相关性最好的光谱曲线特征变量分别为K pv、Kb和GPA1，相关系数均大于0.80。抛物线和线性模型是估测各地类春小麦叶绿素含量的最佳模型，模型R2均超过0.89，预测能力均大于92%。研究结果为以后利用高光谱信息监测小麦长势及产量提供了新思路和方法。

A Comparative Study on Relationship Between New Characteristic Parameters of Spectral Curve and Chlorophyll Content for Spring Wheat at Different Area

New characteristic parameters of spectral curve was used to estimating chlorophyll content for spring wheat at irrigated and dry land, determination of canopy spectral, chlorophyll content, leaf width and seedling height of spring wheat at different area, using the differences of spectral curve peak-valley of spring wheat at different area. Through analyzing the relationship between characteristic parameters of spectral curve reflectance and chlorophyll content, the chlorophyll content monitor models of different area are established respectively and test its precision. The results showed that, correlation had a significantly difference among irrigated land, shady land, sunny and double-sided land at 723 nm, 730 nm and 736 nm pre and post. Before 723 nm, 730 nm and 736 nm, correlation coefficient was negatively related, after 723 nm, 730 nm and 736 nm, correlation coefficient was positively related. Characteristic parameters from green peak, yellow peak region of spectral curve and anles formed of green peak display better correlation with wheat chlorophyll content, irrigated land, shady land, sunny and double-sided land have significant correlation with Kpv, Kb and GPA1 respectively. Correlation coefficients were higher than 0.80. Parabola and linear model was the best model of estimating spring wheat chlorophyll content at different area, model R2 were more than 0.89, and predictive are higher than 92%. The results would provide new ideas and methods to using hyper spectral information for monitoring wheat growth and yield.