Gao Yongnian, Gao Junfeng, Yin Hongbin, Liu Chuansheng, Xia Ting, Wang Jing, Huang Qi
State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 210008, China.
State Key Laboratory of Lake Science and Environment, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing 210008, China.
J Environ Manage. 2015 Mar 15;151:33-43. doi: 10.1016/j.jenvman.2014.11.036. Epub 2014 Dec 19.
Remote sensing has been widely used for ater quality monitoring, but most of these monitoring studies have only focused on a few water quality variables, such as chlorophyll-a, turbidity, and total suspended solids, which have typically been considered optically active variables. Remote sensing presents a challenge in estimating the phosphorus concentration in water. The total phosphorus (TP) in lakes has been estimated from remotely sensed observations, primarily using the simple individual band ratio or their natural logarithm and the statistical regression method based on the field TP data and the spectral reflectance. In this study, we investigated the possibility of establishing a spatial modeling scheme to estimate the TP concentration of a large lake from multi-spectral satellite imagery using band combinations and regional multivariate statistical modeling techniques, and we tested the applicability of the spatial modeling scheme. The results showed that HJ-1A CCD multi-spectral satellite imagery can be used to estimate the TP concentration in a lake. The correlation and regression analysis showed a highly significant positive relationship between the TP concentration and certain remotely sensed combination variables. The proposed modeling scheme had a higher accuracy for the TP concentration estimation in the large lake compared with the traditional individual band ratio method and the whole-lake scale regression-modeling scheme. The TP concentration values showed a clear spatial variability and were high in western Lake Chaohu and relatively low in eastern Lake Chaohu. The northernmost portion, the northeastern coastal zone and the southeastern portion of western Lake Chaohu had the highest TP concentrations, and the other regions had the lowest TP concentration values, except for the coastal zone of eastern Lake Chaohu. These results strongly suggested that the proposed modeling scheme, i.e., the band combinations and the regional multivariate statistical modeling techniques, demonstrated advantages for estimating the TP concentration in a large lake and had a strong potential for universal application for the TP concentration estimation in large lake waters worldwide.
遥感技术已广泛应用于水质监测，但大多数此类监测研究仅关注少数水质变量，如叶绿素a、浊度和总悬浮固体，这些通常被视为光学活性变量。遥感技术在估算水中磷浓度方面存在挑战。湖泊中的总磷（TP）已通过遥感观测进行估算，主要使用简单的单波段比值或其自然对数以及基于现场TP数据和光谱反射率的统计回归方法。在本研究中，我们研究了使用波段组合和区域多元统计建模技术从多光谱卫星图像建立空间建模方案以估算大型湖泊TP浓度的可能性，并测试了该空间建模方案的适用性。结果表明，HJ - 1A CCD多光谱卫星图像可用于估算湖泊中的TP浓度。相关性和回归分析表明，TP浓度与某些遥感组合变量之间存在高度显著的正相关关系。与传统的单波段比值法和全湖尺度回归建模方案相比，所提出的建模方案在大型湖泊TP浓度估算方面具有更高的准确性。TP浓度值呈现出明显的空间变异性，巢湖西部较高，巢湖东部相对较低。巢湖最北部、东北沿海地带和巢湖西部东南部的TP浓度最高，除巢湖东部沿海地带外，其他区域的TP浓度值最低。这些结果有力地表明，所提出的建模方案，即波段组合和区域多元统计建模技术，在估算大型湖泊TP浓度方面具有优势，并且在全球大型湖泊水域TP浓度估算方面具有很强的普遍应用潜力。
