Scaling up Soil Properties for Predicting Crop Yield with DSSAT-CSM Model.

The Decision Support System for Agrotechnology Transfer (DSSAT) comprises crop simulation models (CSM) for over 28 crops and has been in use for more than 20 years by researchers, educators, consultants, extension agents, growers, and policy and decision makers in over 100 countries. Despite its broad application, little is known about the effect of spatial variability and spatial averaging of soil properties on the accuracy of DSSAT-CSM in predicting crop yields. The objective of this work was to examine (a) spatial variability of soybean yields across an agricultural field, (b) how spatial variability in soil properties translates into yield variability, (c) how up-scaling of soil properties affects model accuracy in predicting soybean yields for different weather conditions. The study was conducted at LTER KBS in southwest Michigan, USA. Soybean yield was measured at 2.2 ha and 4.9 ha fields at 2x5 m resolution in 2010. Soil properties were measured at 21 locations, which represented two soil types and 3 topographical elements, i.e. summit, slope and depression. The DSSAT-CSM was calibrated on soybean yields measured at these 21 locations. The calibrated parameters were then used to predict soybean yield for 22 years with varying weather conditions. The model parameters were then scaled up using different techniques, such as averaging soil properties, averaging the model parameters estimated from measured soil properties, and using typical soil profile descriptions and the SSURGO soil database. The results of this study showed a soybean yield varying from 44 to 4500 kg/ha. The calibrated model accurately predicted yield values measured at 21 locations. All up-scaling techniques resulted in an overestimation of soybean yields. Deviations of yield values obtained with up-scaled parameters from those computed using the calibrated model were the highest for (1) up-scaling via using soil data estimated from soil profile description and the SSURGO database, (2) for dry years when the potential evapotranspiration considerably exceeded precipitation and resulted in the highest water stress in soybean plants. Overall, this study indicated that results of DSSAT-CSM using these types of up-scaled parameters should be regarded with caution, especially in dry growing years.