天然气微泄漏胁迫下大豆冠层叶绿素含量的高光谱估测

Hyperspectral Estimation of Canopy Chlorophyll Content in Soybean under Natural Gas Micro-leakage Stress

通过野外实验对天然气微泄漏胁迫下大豆冠层叶绿素含量(Canopy Chlorophyll Content,CCC)进行高光谱估测,进而根据CCC的变化信息辅助判断天然气微泄漏信息。对大豆冠层光谱进行分数阶微分处理,根据各微分光谱与CCC的相关性选取敏感波段,构建各微分阶数下的多元线性回归模型和BP(Back-Propagation)神经网络模型,并对模型进行精度评价,筛选出天然气微泄漏胁迫下大豆CCC高光谱估测的最优模型。通过对各模型的精度评价可知:基于0.5阶微分的BP神经网络模型建模集R^2为0.914、RMSE为0.241g/m^2、RPD为3.351,验证集R^2为0.873、RMSE为0.294g/m^2、RPD为2.465,在所有模型中,其精度最高、预测效果最好,且稳定性好。因此,该模型可用于天然气微泄漏胁迫下大豆CCC的高光谱估测。

With the extensive use of natural gas,its leakage during storage and transportation has become a serious issue,which may significantly impact social economy and safety.Therefore,it is crucial to detect nature gas leakage timely.In this paper,we have reported the results of estimating the canopy chlorophyll content (CCC) of soybean under natural gas micro-leakage stress through canopy spectrum.These results can assist in assessing the micro-leakage of natural gas based on the change of CCC.The canopy spectrum of soybean was first subject to fractional differential processing.Sensitive bands were then selected based on the correlation between each differential spectrum and the CCC.Multiple linear regression models and BP neural network models were constructed.The coefficient of determination,root mean square error and relative prediction ability of these models were calculated and analyzed.The fractional differential spectrum has some advantages over the original spectrum.The best-fit differential order is different for different modeling methods.The accuracy of the BP neural network model is better than the multiple linear regression model in assessing the soybean CCC under natural gas micro-leakage stress.The BP neural network model based on the 0.5-order differential results in the highest precision and the best prediction power among all the models considered in this study.Furthermore,the stability of the model is good.Therefore,it can be used for the hyperspectral estimation of the soybean CCC under natural gas micro-leakage providing a basis for assessing micro-leakage point of natural gas.