中文摘要：

悬浮物浓度是水质评价的一项重要指标，利用遥感技术准确获取区域面状水域悬浮物浓度信息，是遥感监测水质参数的一项重要任务。利用地面实测的14个样点的光谱数据。将2004年7月26日太湖TM数据的DN值，校正为遥感反射率，并利用Gordon模型和太湖水体固有光学特性，建立水体反射率模拟的分析模型，模拟水体R（0^-），进而利用TM数据反演水体悬浮物浓度，并绘制太湖悬浮物浓度分布图。将悬浮物浓度反演的结果与14个样点的实测结果相比较，其中，有79％的样点的估计精度高于70％，有64％的样点的估计精度高于80％。

英文摘要：

Concentration of suspended sediments is an important indicator of water quality evaluation. The method of using remote sensing technology to monitor water quality by obtaining the suspended sediments concentration （SSC） was applied in this paper, and the inversion of SSC from TM image was introduced based on an analytic model. Based on TM image on 26th July 2004, in situ spectra and matching SSC measurement on 27th July 2004 were taken on 14 sampling stations in Meilianghu in northern of Taihu Lake. Water samples were collected and analyzed for SSC and chlorophyll content. The procedure of data process and analysis includes three steps. Firstly, the in situ spectra are converted to remote sensing reflectance, and then a linear regression equation is built to calibrate the DN values of TM image to ground remote sensing reflectance. Secondly, R （0^-） simulation model is built with Gordon model using formerly established IOPs （ Inherent Optical Properties）. RMSE was below 0.05 and the correlation coefficient was above 0.85 for all the 14 samples. Thirdly, the SSCs in Meilianghu were conversed using calibrated TM value. With lsqcurvefit tool of Matlab, TM band 2 and 3 values were input to the reversion model. The conversed image of Meilianghu showed that the maximum SSC was 118g/m^3 , and the minimum SSC was 44g/m^3, in average of 77g/m^3. The SSC ranged from 40 to 60g/m^3 in 20% of the region, from 60 to 80g/m^3 in 28% of the region, 48% of the region varied from 80 to 100g/m^3 , and in 4% of the region, from 100 to 120g/m^3. The accuracy was assessed by in situ values of 14 samples. Comparison between the conversed and measured SSC data showed that 79% of conversed data whose estimation precision were over 70% , and 64% , over 80%. The maximum relative error was 0.41 that was taken near lake bank, and the minimum one was 0.01. In this study, the authors also built an analytical model and used a linear correction method to reduce of atmosphere effect. The high precision of converted data show