Effects of hidden structural uncertainty from pedotransfer functions on crop production, water balance, and nitrogen loss

Résumé fourni par la source

Process-based agroecosystem models are widely used for assessing crop production, ecosystem services, and environmental risks under changing management and climate conditions. Hydrological processes strongly influence plant growth and nutrient dynamics. At large spatial scales, soil hydraulic parameters are typically estimated using pedotransfer functions (PTFs). While previous studies have been focused mainly on hydrological outputs, the effects of PTF choice on crop yields, nitrogen losses, emissions, and agroenvironmental trade-offs remain underexplored. In this study, the impacts of PTF choice on the simulated crop yield, evapotranspiration, nitrate leaching, and N₂O emissions were quantified using the HERMES2Go model at three levels: (i) the experimental plot scale at Polkovice, (ii) the district scale across arable land in the Přerov district, and (iii) a synthetic set of 34 soils representing a broad texture gradient. Nine PTFs were applied across four crops from 2000 to 2022. At the plot scale, no single PTF consistently outperformed the others, and differences among PTFs were limited under site-specific conditions. In contrast, district-scale simulations revealed substantial variability in plant-available water (PAW), ranging from 9.1 to 27.8 mm in the 0–30-cm layer. This variability translated into contrasting yield–PAW relationships, crop-specific PTF rankings, and evapotranspiration differences. Pareto analyses confirmed structured trade-offs between productivity and nitrogen losses, with no PTF being universally optimal across environmental indicators. A synthetic 34-soil experiment revealed that PTF-induced uncertainty was strongly soil dependent and nonlinear across the PAW gradient. Among different soils, the median yield among PTFs was 2.98 t ha⁻¹ (maximum ≈ 8 t ha⁻¹), while the relative variability ranges were 14.3–248.9% for nitrate leaching and 11.8–80.4% for N₂O emissions. In comparison, evapotranspiration was robust (median coefficient of variation (CV) ≈ 3.1%). PAW alone did not explain soil sensitivity to PTF choice. Overall, PTF effects depended on soil properties, crop type, environmental indicators, and spatial scale, demonstrating that PTFs are important sources of structural uncertainty in regional agroenvironmental modeling.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.