Calibração e validação de um modelo de emissão de nós na cultura da soja

Abstract

Soybean (Glycine max L.) is the most cultivated oilseed in the world. Mathematical models adapted and tested under different environmental conditions are useful tools that allow describing the interactions between the plant and the environment, and thus, can be used in studies to better understand the processes of growth and plant development. The objective of this study was to fit a nonlinear model for the simulation of node emission of soybean varieties recommended for the subtropical region in Southern Brazil. Data from 47 experiments conducted in the state of Rio Grande do Sul, during eight growing seasons, evaluating 35 soybean varieties with determined and undetermined growth type and relative maturity group ranging from 3.9 to 8.3. The node emission model was calibrated and validated in three versions, regarding the calculation of f (T) in the SOYDEV model. Version 1 (Tmean) - the daily air temperature (TMED) used to calculate f(T) was obtained by the arithmetic mean between the minimum and the maximum daily air temperature; Version 2 (Tmm) - one f(T) was calculated separately with the minimum daily air temperature and one f(T) with the maximum daily air temperature, the average between f(T) in the NN model was used; Version 3 (Thmean) - the daily air temperature (TMED) used to calculate f(T) was obtained by averaging the hourly minimum and maximum air temperatures. The calibration of the maximum node appearance rate (NARMAX) was performed for the three versions of the model, using the method of minimum squares of the residual between the observed and simulated node numbers. From the cultivar-specific calibration three equations were generated in which NARMAX is a function of GMR: one general (regardless of growth type), one for cultivars with indeterminate growth type and one for cultivars with determined growth type. The performance of the three versions of the model was evaluated with independent data of node number from experiments and field observations, in irrigated and non-irrigated conditions, lowland and upland environments and different plant densities. The average RMSE between observed and simulated number of nodes in the Tmm version and general equation was 1.15 nodes. Soybean knot simulation is improved by calculating f(T) for the minimum daily air temperature and f(T) for the maximum daily air temperature and then averaging f(T ) (Tmm version) compared to using the average daily air temperature (Tmean and Thmean versions).