地理国情，我国重要的基本国情，对社会经济发展，城镇化建设等起着重大的影响作用；地理国情通过对国情要素进行监测与统计分析，得到了反映要素空间分布及演变结果。水系作为基础测绘和地理国情监测的重要要素之一，研究水系提取及水系统计分析就至关重要。本文主要采用三种方法进行水系提取，基于传统的监督分类方法、基于eCognition软件的面向对象分类方法以及利用DEM提取水系的GIS方法。最后，在水系提取的基础上，结合四期的水系专题数据，对西安市周至县水系进行基本统计分析。 基于eCognition的水系提取主要分为两步，第一步，选取最优的分割参数，包括尺度参数，形状参数，光谱参数，对遥感影像进行分割，分割质量的好坏影响着分类精度；第二步，在分割的基础上，利用eCognition软件中的特征知识库对地物特征进行分析，选取合适的对象特征，进行水系提取。 本研究以分辨率为2米的ZY3融合影像为实验区，采用监督分类和面向对象的影像分类对水系进行提取，对提取结果进行比较，得到适用于地理国情监测项目的水系提取方法。探索出了利用DEM数据提取水系的GIS方法。 本研究的创新之处在于，自动化的水系提取方法，极大的提高了项目日常生产效率，减少了人工工作量，为国情监测项目提供了一定的技术参考。

Geographical conditions, China's important basic national conditions, it plays a significant role in social and economic development and urbanization construction; through monitoring and statistical analysis of the national conditions elements, geographical conditions has been reflected in the spatial distribution of elements and evolution results. Because the water is the important element of basic surveying and geography and conditions monitoring, it is essential to study the extraction and statistics and analysis of water. This paper mainly use three methods to extract water, traditional supervised classification method, object-oriented classification method based on eCognition software and GIS method for extracting water by DEM. Finally, based on the extraction of water and combined with four special water data, do the basic statistical analysis of Xi'an Zhouzhi county water. The extraction of water system based on eCognition is mainly divided into two steps, First is to select the optimal segmentation parameters, including scale parameters, shape parameters, spectral parameters, and segment the remote sensing images, and the quality of segmentation affects the classification accuracy. Second, on the basis of segmentation, the features of the eCognition software are used to analyze the features of the objects, and the appropriate object features are selected to extract the water system. In this study, ZY3 fusion images with a resolution of 2 meters were used as the experimental area. The water system was extracted by supervised classification and object-oriented image classification. The extraction results were compared and the water extraction methods were applied to the monitoring project. The GIS method of extracting water system by using DEM data is explored. The innovation of this research is that the automatic water extraction method greatly improves the daily production efficiency of the project, reduces the manual workload, and provides some technical reference for the national condition monitoring project.