Skip to Main Content
In this study, we evaluate the benefits of the use of broad-band reflectance data from remote-sensing imagery for improving point prediction of N (Nitrogen) and PC (Phosphorus Concentration) in lakes. The algorithms were calibrated and validated by in situ measurements, collected on 27 and 28 October, 2003, in Taihu Lake, China. Both two algorithms produced well performance in estimating NC and PC in Taihu Lake, but the PC retrieval model had a superior performance to NC retrieval model. The RE (Relative Error) of the PC and NC retrieval models were 11.7% and 35.6%, respectively. According to no more than 30% accuracy requirements of water quality estimation for remote-sensing technology, the accuracy of PC retrieval model is more acceptable than the NC retrieval mode's. Finally, the PC and NC were estimated from Landsat/TM imagery, collected on 28 October, 2003. The retrieval results showed that the NC and PC were higher in the south, east and center of the lake ( >;18 mg/l and >;2 mg/l for NC and PC, respectively) and lower in the west and north of the lake ( <;18 mg/l and <;2 mg/l for NC and PC, respectively). Although it was a special case study of this paper, the modeling procedures of NC and PC estimation algorithms could give us the implication when we used the remote-sensing technology to estimate the NC and PC from similar or dissimilar aquatic environments.