摘要
以太湖及洞庭湖为例,检验海洋一类水体颗粒有机碳(particulate organic carbon,POC)浓度遥感估算方法在内陆湖泊二类水体中的适用性,结果表明,一类水体POC反演模型并不适用于二类水体。针对二类水体建立了以近红外波段(834nm)为因子的单波段POC反演模型以563及834nm波段组合为因子的两种双波段反演模型,模型验证结果显示,单波段模型的均方根误差(RMSE)为1.12mg/L,平均相对误差(MAPE)为35.8%,两个双波段反演模型的RMSE分别为1.09mg/L及1.11mg/L,MAPE分别为37.3%及37.8%,三种模型均可用于太湖及洞庭湖水体的POC浓度遥感估算。在此基础上,以太湖为例,建立了基于静止轨道卫星海洋水色(GOCI)卫星数据的太湖POC反演模型,反演模型的MAPE为35%。利用5月13日8景GOCI影像,研究了太湖POC浓度日变化,发现POC浓度日变化存在两个阶段:上午至中午的递减阶段和中午至傍晚的递增阶段。
A deep understanding of particulate organic carbon is of great significance for the further research of the carbon cycle of water.After testing the applicability of ocean inversion models of POC in inland lakes,results showed that these ocean models cannot be applied in Lake Taihu and Lake Dongting.So we established a single band inversion model and two kinds of dual-band inversion model to be used in case two water,by the use of a near-infrared band（834nm）and 563 nm with 834 nm respectively.The validation of the models shows that the RMSE of the single band model is 1.12mg/L,MAPE is 35.8%.The RMSE of two kinds of dual-band semi-analytical inversion models are 1.09mg/L and 1.11mg/L respectively,and MAPE are 37.3% and 37.8%.All three kinds of models can be used for estimating the concentration of POC in Lake Taihu and Lake Dongting.In order to use remote sensing to observe the variation of POC concentration,we established a model that can be used on GOCI in the region of Lake Taihu,the MAPE of this model is 35%.Then we estimated the POC concentration distribution of the whole lake using GOCI satellite data.Combining the characteristics of GOCI data with high temporal resolution,we analyzed the POC daily variation,and found that the concentration of POC increased from morning to noon,and declined from noon to evening.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2015年第12期1618-1624,共7页
Geomatics and Information Science of Wuhan University
基金
国家自然科学基金资助项目(41171269)~~
