Survey of indicators for the elaboration of a conceptual model of soil radio-vulnerability to support waste management actions and remediation of iodine contaminated areas

Computational radioecological models used for environmental safety purposes depend on parameter values not always available with the required environmental specificity. Within the agricultural scenario, the soil-plant transfer factor (Fv) is a crucial information for the evaluation of the radiological risk due to the ingestion of food. The soil-plant transfer is directly related to plant phenology and physiology, soil properties and physicochemical properties of radionuclides. The objective of this work was to identify, through the systematic review of the literature, the soil properties that affect iodine Fv values and, from these, to propose its potential soil indicators to forecast Fv values by artificial intelligence tools. The literature review allowed to select the organic matter content and the pH of the soils as potential soil indicators of Fv values. Despite the abundant literature on the subject, only 125 Fv values for iodine could be associated with at least one soil property that were collected and structured in a database. From this database, only 2 learning sets were structured for training of an artificial neural network and, only for ¹²⁵I: one for wheat (n = 38) and one for lettuce (n = 14). This result indicates that the records of Fv values do not report the adequate environmental information for the analysis of this parameter, and all information was restricted to temperate or subtropical climate soils. The artificial neural network trained with these results, was able to provide sufficiency of organic matter content and the pH soil attributes to forecast soil-plant transfer factor values to Iodine validating the used methodological approach to use AI tools to predict soil-plant Fv values. (author)