Obtaining accurate scientific observation and identification data is crucial to establishing a comprehensive layout of crop germplasm resource observation and identification sites. The purpose of this research was to establish a rationally distributed, normative and scientific long-term observation and identification system for germplasm resources, carry out comprehensive identification and evaluation of the importance of resources, integrate observation and identification data, provide necessary content for the construction of a crop germplasm resources big data system and provide solid basic data support for agricultural scientific research and modern seed industry development.

First, 379 sites were identified on the basis of the investigation. Then the environmental factors most closely related to crop growth were determined, and the factor values of each site were calculated using the climate production potential. Finally, these factors were used to select the best method for identifying spatial clustering, and generate a Tyson polygon for each type of site to create area divisions.

With consideration of the above methods, it was determined that temperature, precipitation, altitude, latitude and sunshine duration were the environmental factors most closely related to crop growth. The stations were clustered into 26 categories, China was divided into 26 evaluation and identification areas by generating a Tyson polygon, and six site layout principles were set on the basis of zoning to guide site selection in the region.

The layout of the station system affects the representativeness and scientific accuracy of future observation and evaluation results. The selection of stations is a highly complex process; the site system layout method used in this research effectively combined environmental and spatial factors, which not only effectively guided the selection of multiple environmental assessment sites in the crop germplasm resources field, but also provided reference for the layout of observation sites in crop-related fields, so as to obtain more valuable and representative identification data.