Zhang Dong, Xu Yong, Zhang Ying, Li Huan
School of Geographic Science, Nanjing Normal University, Nanjing 210046, China.
Huan Jing Ke Xue. 2010 Jun;31(6):1435-41.
Choosing dissolved inorganic nitrogen (DIN) as one of the representative nutritional salt monitoring indexes, a hyperspectral remotely sensed inversion model was built and applied to quantitatively retrieve water quality parameters with its spatial distribution patterns in coastal estuary with high suspended sediment concentration (SSC). It was found that when SSC was larger than 0.1 kg/m3, DIN concentration had a notable inverse correlation with SSC and the correlation coefficient R2 reached 0.617. Based on this conclusion, firstly the in-situ observed water surface remote sensing reflectance was resampled according to the spectral response characters of Hyperion sensor. And then, statistical correlation analysis between reflectance and DIN concentration was carried out. The results showed that band reflectance of R804 and R630 representing the second and first reflectance peak of water spectrum curve were sensitive to the variation of DIN concentration. And then, a pseudo remotely sensed sand parameter index R804 x R630/(R804 - R630) was calculated for the construction of the nonlinear DIN quantitative reversion model. Correlation coefficient R2 between observed and simulated DIN concentrations for 29 calibrating samples and 10 validating samples were 0.746 and 0.67, while their mean absolute errors reached 109.07 and 147.58 microg/L, respectively. The model was then applied on Hyperion hyperspectral image to get the spatial distribution character of DIN concentration in Sheyanghe river estuary and the DIN concentration was between 52 to 513 microg/L. Results indicated that in coastal estuary which was dominated by suspended sediments, the diffusive trends of DIN concentration reversed by remote sensing techniques had an intimate relationship with motions of tidal current and transportation attributes of SSC. As the hydrodynamic conditions were unclear, hyperspectral remote sensing technique was an effective technical way for dynamic survey of DIN concentration.
选择溶解无机氮（DIN）作为代表性营养盐监测指标之一，构建了高光谱遥感反演模型，并将其应用于高悬浮泥沙浓度（SSC）的海岸河口水质参数及其空间分布格局的定量反演。研究发现，当SSC大于0.1kg/m³时，DIN浓度与SSC呈显著负相关，相关系数R²达到0.617。基于这一结论，首先根据Hyperion传感器的光谱响应特征对实测水面遥感反射率进行重采样。然后，开展反射率与DIN浓度之间的统计相关分析。结果表明，代表水光谱曲线第二和第一反射峰的R804和R630波段反射率对DIN浓度变化敏感。进而，计算了伪遥感沙参数指数R804×R630/（R804 - R630），用于构建非线性DIN定量反演模型。29个校准样本和10个验证样本的实测与模拟DIN浓度之间的相关系数R²分别为0.746和0.67，而它们的平均绝对误差分别达到109.07和147.58μg/L。然后将该模型应用于Hyperion高光谱图像，得到射阳河口DIN浓度的空间分布特征，DIN浓度在52至513μg/L之间。结果表明，在以悬浮泥沙为主的海岸河口，遥感技术反演的DIN浓度扩散趋势与潮流运动和SSC的输运特性密切相关。由于水动力条件不明，高光谱遥感技术是DIN浓度动态监测的有效技术手段。
