全国地理国情普查是我国的基本国策，该项调查主要是通过采用各种测绘技术，对我国地理信息资源进行科学、准确的调查。该项调查以自然、生态、资源、环境、人类活动为基本调查对象，普查结果可以用于各领域的地理信息服务。地理要素包括诸多类型，其中地表植被是最基础的类型，主要用于对涉及生态环境及资源类研究，是一种基础又重要的衡量指标。 我国的地理国情普查工作中以多源遥感影像为数据。但在日常工作中发现遥感数据由于多源性、多时相性很大程度上会对地表覆盖分类体系中高、中、低覆盖度的草地覆盖度的估算结果质量造成影响。目前，对于三类草地覆盖度估算和划分一般采用人工经验加外业验证的方法，但此方法受限于人为因素和数据质量，会产生分类精度低、工作效率低、耗时长等问题。本文以青海省海北藏族自治州为研究区域，基于LandSat8中分辨率卫星影像，通过计算反演植被指数，提取植被信息，针对遥感影像建立二分模型，结合野外实地采样结果，对一定范围内的草地进行高、中、低覆盖度草地进行覆盖度估算，最后基于估算结果进行草地类别自动划分。实验最后将估算结果与外业实测数据进行了对比验证，结果表明本文方法较好的对实验区域三类草地进行了划分，覆盖度估算结果与外业实测值的统计误差为0.06，三类草地划分正确率最高为100%，最低为74.82%，研究区域内高、中、低覆盖度草地面积分别为18026.17km2、1775.11 km2、678.88 km2。研究证明了该方法的可行性，并且可以较大程度的保障地理国情普查工作的进度，减轻了人工解译植被的工作量、缩短了解译工作时间。本文的研究方法及数据对研究我国高原草地生态具有一定的科研意义，对地理国情普查中三类天然草地覆盖度估算的实际工作具有一定的实用意义。

The national census of geographical national conditions is the basic national policy of China. The investigation is mainly based on the use of various surveying and mapping techniques to conduct scientific and accurate surveys on China's geographical information resources.The survey focused on nature, ecology, resources, environment, and human activities. The results of the census can be used for geographic information services in various fields. Geographical elements include many types, of which the ground vegetation is the most basic type. It is mainly used for research involving the ecological environment and resources and is a basic and important measure. The multi-source remote sensing image is used as the data in our country's geographic national survey work.In daily work, it is found that the multi-source and multi-temporal nature of remote sensing data will largely affect the quality of the estimation of the coverage of grassland with high, medium, and low coverage in the land cover classification system.At present, the method of artificial experience plus field verification is generally used to estimate and classify the coverage of the three types of grassland. However, this method is limited by human factors and data nature, which may result in problems such as low classification accuracy, low work efficiency, and long time-consuming.This paper takes Haibei Tibetan Autonomous Prefecture of Qinghai Province as the study area, based on the LandSat8 medium-resolution satellite image, calculates the inversion vegetation index, extracts the vegetation information, sets up a binary model for the remote sensing image, and combines field field sampling results to perform the grassland within a certain range. The coverage of high, medium and low coverage grassland was estimated, and the grassland category was automatically divided based on the estimation results.At the end of the experiment, the estimation results were compared with field measurement data. The results show that the method has better classification of the three types of grassland in the experimental area. The statistical error between the coverage estimation result and the field measurement value is 0.06. The correct rate of grassland division is 100% and the lowest is 74.82%. The grassland area with high, medium and low coverage in the study area is respectively 18026.17km2, 1775.11km2, and 678.88km2.The study has proved the feasibility of the method, and can greatly guarantee the progress of the census work of the geographic national conditions, reduce the workload of manual interpretation of vegetation, shorten the understanding of the translation time, and cover the three types of natural grassland in the geographical national survey. The estimation work has certain practical significance.