This paper is the result of the soil geochemical survey engineering.
[Objective] Zinc (Zn) is an essential trace element for human body. Using regional geochemical survey data to accurately predict the Zn content in crops and carry out development planning for Zn-rich agricultural products is still a problem. [Methods] In this paper, we chose Linshui County of Sichuan Province as the study area. Basing on the investigation results obtained from the geochemical survey of land quality, content and spatial distribution characteristics of Zn in the soil and crop were studied and the factors affecting Zn element uptake by maize and rice grain were analyzed. [Results] The Zn content of topsoil in the study area ranged from 25.00-142.00mg/kg with a mean value of 81.93 mg/kg. The higher content of Zn in soil were mainly distributed in exposure of carbonate rocks and Emei Shan basalt in Huaying mountain. The average content of Zn in maize and rice were 17.18 mg/kg and 11.20 mg/kg, respectively. The Zn enrichment rates were 44.0% and 8.2%, respectively. The prediction of the planting areas of Zn rich maize and Zn rich rice in Linshui County reached 235.34 km² and 30.99 km² respectively by using back-propagation neural network models. [Conclusions] The main factors affecting the Zn accumulation of maize and rice in the study area were Fe₂O₃, Mn, pH, SiO₂/Al₂O₃, Cao, organic matter and nutrient element P in soil. The back-propagation neural network models could better simulate the relationship between Zn in crop grains and physicochemical properties of soil, which could be used for region specific calculation of crop Zn content.
Highlights: Zn content in maize and rice grains is not only affected by soil Zn but major elements in soil; A map of predictive maize and rice Zn is proposed by neural network models.