Soils in Crop Modeling: Is Uncertainty in Soil Data Higher Than Climate Change Signal.

Crop models have been used to simulate the effect of projected climate change effects on agriculture. Soil data is usually considered as a fixed and static input where one or few soil profile descriptions are used to represent from plot to wide regions. Many publications have demonstrated that field soil sampling and laboratory measurements can generate certain variability in soil parameter values, which added to the inherent soil spatial variability can accumulate a large amount of uncertainty with large effects on the crop model simulations. Usually, a soil profile description is defined as a set of absolute deterministic values for each soil parameter at different soil horizons without taking into account the soil variability when in reality this is not true. In this paper we developed a framework that incorporates all the potential variability on each soil parameter and use the Decision Support System for Agrotechnology Transfer (DSSAT) as the crop model to measure the uncertainty produced by the lack of soil variability information. As a study cases we used two soil profiles, one for Henry County (Typic Hapludults) in Alabama where peanut is produced, and another in Fillmore County (Typic Argialbolls), Nebraska,  where corn is produced. Using the original deterministic soil profile descriptions and by means of the delta method, we also simulated the effects of climate change projections on crop production. We then compared the uncertainties on simulated crop yield of projected climate change scenarios with its counterpart produced by not considering soil variability. The objective of this paper is to compare the uncertainty due to the variability in the soil profile data versus the uncertainty and magnitude due to climate change projections in simulated yields. The effects of variability of specific soil parameters were analyzed.