基于遥感影像的图像类推法提高湖泊水量监测精度的研究

发布时间：2018-02-28 23:33

本文关键词： 图像类推水体插值法遥感影像库容　出处：《西南交通大学》2014年硕士论文　论文类型：学位论文

【摘要】：湖泊不仅是城市水资源的主体,还是农业灌溉的重要水源。然而,由于气候的变化及人类的过度开发,导致湖泊生态环境受到严重破坏,水量急剧萎缩。因此,对湖泊水量变化的实时监测将会对城市、农业用水起到重要的作用。 Landsat卫星遥感影像具有范围大、周期短、现势性强、获取免费等特点,在湖泊水量监测方面具有不可比拟的优势。然而,TM影像的空间分辨率仅可达到15m或30m,这就会给水量的实时监测带来很大的误差。国内针对TM影像进行湖泊库容监测方面已经有了一定的研究,但如何提高监测精度方面还涉及较少。因此,在低空间分辨率影像的前提下提高湖泊水量的监测精度将成为研究的关键。鉴于此,本文开展基于图像类推法的湖泊库容提取研究。首先,联合图像类推原理与立方卷积插值法生成超分辨率影像,从而有效提高TM影像的空间分辨率。然后,通过分析超分辨率影像中水体的特性,开展水体信息提取试验。分析坡度与高差的关系,通过在湖泊坡度平缓地区的水位变化来监测水量变化,将水位的垂直变化转换为水平方向的变化,并利用区域内插法提取水位线信息。最后,通过结合由湖泊等深线所生成的DEM模型,有效获取湖泊的库容信息。论文以Lake Jackson为实验对象,利用该方法对多期影像进行试验,结果表明,基于图像类推法的湖泊水位精度平均提高了0.223英尺,库容精度平均提高了3481.2万立方英尺。该方法可以在相同空间分辨率影像的条件下提高湖泊水量的监测精度,对湖泊库容量的评估更为精确,将会对水资源分配、利用以及政府的管理和决策起到一定的现实意义。
[Abstract]:Lake is not only the main body of urban water resources, but also an important water source for agricultural irrigation. However, due to climate change and over-exploitation of human beings, the ecological environment of lakes is seriously damaged and the water volume shrinks sharply. Real-time monitoring of lake water change will play an important role in urban and agricultural water use. Landsat satellite remote sensing images have the characteristics of large range, short period, strong reality and free access. However, the spatial resolution of TM image can only reach 15m or 30m, which will bring great error to real-time monitoring of water quantity. There has been some research in the field of measurement. However, how to improve the monitoring accuracy is less concerned. Therefore, improving the monitoring accuracy of lake water quantity under the premise of low spatial resolution image will be the key to the research. In view of this, this paper studies the extraction of lake storage capacity based on image analogy. Firstly, combining the principle of image analogy with cubic convolution interpolation method to generate super-resolution image, thus effectively improve the spatial resolution of TM image. By analyzing the characteristics of water body in super-resolution image, the experiment of extracting water information is carried out, the relationship between slope and elevation difference is analyzed, and the change of water level is monitored through the change of water level in the area with gentle gradient of lake. The vertical change of water level is transformed into the horizontal change, and the information of water level is extracted by region interpolation. Finally, by combining the DEM model generated by the lake isobath, Taking Lake Jackson as the experimental object, the multi-period image is tested by this method. The results show that the accuracy of lake water level based on image analogy method is improved by 0.223 feet on average. The accuracy of reservoir capacity is improved by an average of thirty-four million eight hundred and twelve thousand cubic feet. This method can improve the monitoring accuracy of lake water under the condition of the same spatial resolution image, and the estimation of lake capacity is more accurate, and the allocation of water resources will be improved. Use and government management and decision-making play a certain practical significance.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P237;P332

【参考文献】