珠海一号高光谱卫星的于桥水库水质参数反演初步研究

水体中的悬浮物浓度和透明度是水质调查中的基本参量，也是评价水环境优劣的重要指标。卫星遥感技术可以大范围、较快速、低成本地实现主要水质参数的反演，但是对于水质复杂多变且水域面积较小的内陆水体来说，目前常用的卫星不能满足水质参数的反演需要。珠海一号卫星星座是由珠海欧比特宇航公司发射的新型高光谱卫星星座，具有高空间、高光谱、高时间分辨率的特点，在内陆水体水质监测中具有重要潜力；然而目前对于运用珠海一号数据进行内陆水体水质参数遥感监测的适用性有待研究，需要开展星地同步实验检验其应用效果。以珠海一号高光谱卫星作为遥感数据源，以于桥水库为研究区，利用与卫星同步的水面实测数据，使用经验回归的方法分别构建了悬浮物和透明度的反演模型，并开展了悬浮物浓度和透明度的模型精度检验。检验结果证明，于桥水库悬浮物浓度和透明度的反演模型最佳波段组合分别为R_rs(684)/R_rs(540)和R_rs(656)/R_rs(556)，模型反演相对误差分别为8.6%和11.7%，均方根误差分别为1.0 mg·L-1和18.2 cm。利用建模公式得到了2018年11月22日于桥水库的水质参数分布图，发现于桥水库的悬浮物浓度空间分布呈现北高南低的特点，而水体透明度的空间分布呈现出北低南高的分布特征，这种空间分布特征应该主要是因为于桥水库北部水较浅南部水较深，水越深底泥越不容易再悬浮导致的。基于星地同步实验的于桥水库水质参数反演建模和检验结果初步说明了珠海一号高光谱卫星在定量反演内陆水体水质参数方面具有潜力，未来需要开展更多的星地同步实验进一步完善预处理和反演模型。

Preliminary Study on Water Quality Parameter Inversion for the Yuqiao Reservoir Based on Zhuhai-1 Hyperspectral Satellite Data

The concentration of suspended matter and water clarity are basic parameters for water quality investigations and important indexes to evaluate the quality of water environments.Remote sensing technologies can be used to realize the inversion of the main water quality parameters in a large range,quickly,and at low cost;however,those same technologies cannot be applied to inland water bodies that extend over small areas and have complex and variable water qualities.The Zhuhai-1 satellite constellation is composed of new hyperspectral satellites launched by the Zhuhai Orbita Aerospace Technology Company.Such satellites possess high spatial,temporal,and hyperspectral resolutions;moreover,they have considerable potential for the monitoring of inland water quality.The applicability of remote sensing to the monitoring of inland water quality parameters using data from the Zhuhai-1 satellites has not been studied in detail,and satellite ground synchronization experiments need to be carried out to test its effectiveness.In this paper,we considered remote sensing data collected from those hyperspectral satellites to study the Yuqiao reservoir:based on water surface synchronous data,we constructed inversion models of suspended matter and water clarity using the empirical regression method,and subsequently tested their accuracy.The test results showed that the best band combinations for the inversion models,in order to obtain the concentration of suspended matter and the water clarity in the Yuqiao reservoir,were R rs(684)/R rs(540)(relative error=8.6%;root mean square error=1.0 mg·L^-1)and R rs(656)/R rs(556)(relative error=11.7%;root mean square error=18.2 cm),respectively.The distribution map of the water quality parameters in the Yuqiao Reservoir on November 22,2018 was obtained by using the modeling formula:suspended matter was found to be more abundant in the north than in the south,while water clarity was lower in the north and higher in the south.These spatial distributions mainly depended on the configuration of the Yuqiao reservoir,in which the water depth increases from north to south:in the case of a deeper water column,sediments are less likely to be resuspended.According to the inversion modeling of the water quality parameters and to the test results of the measurements conducted on the Yuqiao Reservoir,we conclude that the hyperspectral remote sensing data of Zhuhai-1 can be used to retrieve the water quality parameters quantitatively.These results have important implications for the monitoring of inland water quality;still,more satellite-ground synchronization experiments are needed to improve further the preprocessing of data and the correspondent inversion models.