Caracterização das regiões produtoras de feijão-comum baseada no risco climático
Nenhuma Miniatura disponível
Data
2024-02-28
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Goiás
Resumo
The aim of this study was to define the climatic risks of yield loss in common
bean and identify and classify the regions producing this grain in Brazil, through computer
simulation, functional data analysis and machine learning tools. Simulations were carried
out with the CSM-CROPGRO-Dry Bean model for different municipalities, sowing dates
and growing seasons (wet, dry and winter). To define the yield loss, the loss curves were
calculated by season, year and municipality, based on simulated yield. Subsequently, the
curves were grouped using functional K-means and a functional average loss curve was
established to explain the behavior of each group of municipalities. To identify and classify
homogeneous common bean production regions, K-means and decision tree machine
learning techniques were used to, respectively, group homogeneous environments based on
simulated yield and associate environmental covariates with homogeneous environments. In
general, during the wet season, the delay in sowing contributed to the increase in common
bean yield. At later dates (between 10/dec and 30/dec) yield loss of less than 20% were
observed. During the dry season, the delay in sowing caused a reduction in yield, mainly in
the Central-West and Southeast regions, with the loss exceeding 70% (10/mar). At these
seasons, water stress was the main factor in yield losses. In winter, yield varied less markedly
between sowing dates. The use of irrigation at this time provides smaller loss in yield (<
20%), although air temperature is a limiting factor for the expansion of common bean
cultivation.
Descrição
Citação
JUSTINO, L.F. Caracterização das regiões produtoras de feijão-comum baseada no risco climático. 2024. 132 f. Tese (Doutorado em Agronomia: Produção Vegetal) – Escola de Agronomia, Universidade Federal de Goiás, Goiânia, 2024