水体中总氮含量是表征湖泊水质的主要指标, 利用遥感技术对其动态定量监测可以及时掌握湖泊污染状况。 文章以巢湖为例, 利用HJ-1A卫星HSI超光谱遥感数据, 通过分析总氮与叶绿素a、 悬浮物的相关性, 采用回归克里格方法建立总氮浓度的定量反演模型, 实现了对巢湖水体总氮浓度的反演。 结果显示, 波段B72, B79和B97的多元线性组合与总氮浓度的相关系数R2为0.76, 而利用多元回归克里格模型, 相关系数R2提高到0.83。 通过使用这种对常规回归模型残差优化的方法, 能有效提高反演的精度, 为建立通用的湖泊总氮浓度定量反演模型提供了有益的探索。

The content of total nitrogen in the waters is an important index to measure lake water quality, and the technique of remote sensing plays a large role in quantitatively monitoring the dynamic change and timely grasping the status of lake pollution. Taking Chaohu as an example, quantitative inversion models of total nitrogen were established by multivariable regression Kriging under analyzing of an correlation between total nitrogen and chlorophyll-a or suspended solids by HIS hyperspectral remote sensing data of HJ-1A satellite. The result shows that the correlation of 0.76 was discovered between total nitrogen and the multiple combination with band 72, band 79 and band 97, while the correlation could be increased to 0.83 by applying combined model of multiple linear regression and ordinary Kriging. The optimization of the residuals of the conventional regression model can improve the accuracy of the inversion effectively. These results also provide useful exploration for further establishing a common model of quantitative inversion of lake total nitrogen concentration.