Predicting on-farm soybean yields in the pampas using CROPGRO-soybean

MERCAU, J.L., DARDANELLI, J.L., COLLINO, D.J., ANDRIANI, J.M., IRIGOYEN, A. E.H. SATORRE

FIELD CROPS RESEARCH

Año: 2007 vol. 100 p. 200 - 209

0378-4290

Abstract Soybean is the main rainfed crop in a wide range of latitudes and sowing dates of the Argentine Pampas. It is sown alone or as a second crop after other winter and summer crops. Modelling approaches have proved to be helpful in the decision making process. The on-farm evaluation of CROPGRO is rather difficult since input data are scarce and frequently of worse quality than those from experimental works. Moreover, CROPGRO simulation of water dynamic processes and their relation with biomass production has not been comprehensively evaluated in soybean crops. The aims of this study were (i) to evaluate the CROPGRO-soybean performance, with emphasis on water demand and supply and biomass production under water limited conditions, (ii) to generate a revised CROPGRO model improving those aspects, and (iii) to compare simulations outputs using the original and the revised CROPGRO models, with on-farm crop data set. In the revised model, we multiplied potential evapotranspiration by 1–1.22 when LAI increased from 0 to 4.0.We set a root extension rate of 4.0 cm/thermal day and a maximum rooting depth of 2.5 m. Finally, we included a nonlinear equation to simulate the relationship between relative transpiration and relative gross photosynthesis. The ability of the revised CROPGRO-soybean to simulate water content depletion and biomass production was tested against several experiments with an imposed drought period.We also calibrated cultivar parameters using ‘‘ad hoc’’ tests in a range of environments (combinations of sowing dates and locations). The models were evaluated with data from 155 commercial farms. V (%) (root mean square error as percentage of the observed mean) for the total cycle length, vegetative period, and reproductive phase simulations were 7, 13 and 15%, respectively. The revised CROPGRO was more accurate in simulating crop yield, biomass, harvest index and yield numeric components. V (%) values ranged from 11 to 17% (revised version) and from 13 to 22% (original version). Besides, V (%) values for yield were 16% with the revised model versus 32% with the original one, considering only paddocks with higher water stress level. The robust prediction of phenology, biomass and yield components obtained with the revised model across different environmental conditions, support its use in the decision making process of the soybean crop at the farm scale.V (%) (root mean square error as percentage of the observed mean) for the total cycle length, vegetative period, and reproductive phase simulations were 7, 13 and 15%, respectively. The revised CROPGRO was more accurate in simulating crop yield, biomass, harvest index and yield numeric components. V (%) values ranged from 11 to 17% (revised version) and from 13 to 22% (original version). Besides, V (%) values for yield were 16% with the revised model versus 32% with the original one, considering only paddocks with higher water stress level. The robust prediction of phenology, biomass and yield components obtained with the revised model across different environmental conditions, support its use in the decision making process of the soybean crop at the farm scale.V (%) values ranged from 11 to 17% (revised version) and from 13 to 22% (original version). Besides, V (%) values for yield were 16% with the revised model versus 32% with the original one, considering only paddocks with higher water stress level. The robust prediction of phenology, biomass and yield components obtained with the revised model across different environmental conditions, support its use in the decision making process of the soybean crop at the farm scale.V (%) values for yield were 16% with the revised model versus 32% with the original one, considering only paddocks with higher water stress level. The robust prediction of phenology, biomass and yield components obtained with the revised model across different environmental conditions, support its use in the decision making process of the soybean crop at the farm scale. # 2006 Published by Elsevier B.V.2006 Published by Elsevier B.V. Keywords: Soybean; Crop simulation; Water stress; CROPGRO; On-farm evaluationSoybean; Crop simulation; Water stress; CROPGRO; On-farm evaluation