Accurate, reliable and timely acquisition of data for spatial planning of land are prerequisite for the "dual evaluation" and "three types of spatial planning" division and pattern optimization, and the preparation of land and spatial planning. Current data of land surveys, urban and rural land usage, and various geographical classification systems have different classification standards and many data sources, so it is urgent to optimize of the land-use classification system, construct land classification conversion rules and identification methods based on multi-source data. This paper preliminary investigates the territorial planning and land classification system using the inductive comparison method. Compared with the former planning, the latest national territory planning system comprise by major functional area, land utilization and urban rural land planning. As the identification of territory classification is the basic data for land geo-spatial planning. This paper takes Hebi, Henan for example to construct the terrestrial spatial classification system and identifying terrestrial space types based on multi-source data. Hebi has large potential improvement and optimization in terms of ecological transformation and development motivation. Combined with the third national investigation of the territory classification, inductive comparison method is used to explore the classification system of territorial space and land transformation and convergence rules of the existing spatial classification system to achieve effective consecutive connection between municipal scale and multiple classification system based on the comprehensive comparative analysis method. Steps are as follows : firstly, unified coordinate system and attribute structure for multi-source data; secondly, through the spatial superposition analysis of GIS, the spatial superposition analysis of unified good data is carried out, and the reference unit of land class identification under multi-source data is constructed, so that each polygon unit has multi-source data attribute structure; thirdly, the criterion of national land class identification is constructed to solve the problem of land classification conversion conflict of multi-source data; finally, the convergence criterion and combination recognition of land classification are modified based on remote sensing image data. The results show that the land use spatial classification system is divided into four first-level categories: ecological protection land, land for agricultural production and living, land for urban development and regional facility land, 12 second-level categories and 45 third-level categories. This paper establishes the criteria of land class connection conversion and the method of fusion identification, and completes the identification of land classification for the land space planning based on the multi-source data. In 2017, the total land area of Hebi City is 2140.43 km2, the area of ecological protection land, land for agricultural production and living, land for urban development and the facility land are 453.95, 1531.53, 121.06 and 33.89 km2, respectively. This paper provides data support for Hebi City to carry out "ecological/agricultural/urban three types of space" division and pattern optimization, "dual evaluation", and planning.