Monitoring of suspended substance concentration in Three Gorges dam based on the semi- analytical model

Suspended substance (SS) concentration is an important index of water quality evaluation. It is significant to build a quantitative remote sensing retrieval algorithm to inverse the SS concentration. With focus on the problems that empirical method is lack of the support of the physical theory and not robust in quantitative remote sensing monitor of SS concentration in inland water, the feasibility of spectrum and inherent optical properties analysis with clear physical meaning is discussed. Based on the experiment taken on 17^th November 2007, the characteristic spectrum, absorption coefficient and SS concentration are taken on 30 samples distributed along the reservoir where is far away from the dam above 3 kilometers. The absorption characteristic of the suspended substance and the contribution of the main constituents of the water (chlorophyll, suspended substance, DOC) to the total absorption are analyzed. Based on the achievements we have acquired, a semi-analytical model using Gordon model is built to inverse the SS concentration. With the SS concentration on samples, the precision of the inversion model is tested. The results show that the error of SS concentration retrieved can satisfy the later research requirement; and the minimum retrieval error is -6.22%, the maximum retrieval error is 62.83%, the mean retrieval error is 25.65%. Further, the inversion method is applied on the multi-temporal image CHRIS to check the precision of the model, and the mean retrieval error is 21.81%.