Abstract:Selenium (Se) is one of the essential trace elements for humans and animals, and in order to scientifically predict the Se content of crops and realize the rational development and utilization of selenium- rich land resources.Methods: This study systematically collected and analyzed 13042 surface soil samples, 313 sets of maize, 164 sets of potatoes and their corresponding root soil samples, studied the distribution characteristics and influencing factors of soil Se content in the area, established the BP neural network prediction model of Se content in edible parts of potatoes and maize, respectively, and made a rational planning of Se rich cropping areas. The BP neural network prediction model for Se content of potato and maize was established, and the rational planning of Se rich cropping areas was carried out. Results: The mean value of soil Se content was 0.164 μg/g with uneven spatial distribution, and the selenium- rich areas were found in the Qingshui River Plain area in the study area and distributed in a row. The soil Se content in the study area was mainly controlled by its soil- forming parent material, weathering and denudation of rocks, migration with river water, and agricutural irrigation, accopanied by an increase in organic matter content, resulting in an enrichment of Se in Quaternay alluvial and diluvial plain. The Se enrichment rates of potato and maize seeds in the study area were 82.32 % and 38.02 %, respectively, and the heavy metal content did not exceed the standard, which had the potential to develop selenium- rich agricultural products. Conclusions: The Se content of crop seeds was mainly related to the Se, S, pH and organic matter content in the root soil, and the Se rich potato planting area in the study area was planned to be 1050.11 km2 and the Se rich maize planting area was 19.19 km2 by the prediction model of Se content of crop seeds.