Full text
| |
| Author(s): |
Lucas Eduardo de Oliveira Aparecido
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Jaboticabal. 2016-04-07. |
| Institution: | Universidade Estadual Paulista (Unesp). Faculdade de Ciências Agrárias e Veterinárias. Jaboticabal |
| Defense date: | 2016-02-19 |
| Advisor: | Glauco de Souza Rolim |
| Abstract | |
The agrometeorological models for crop yield forecasting are robust for strategic planning techniques, since, prior knowledge of agriculture crops facilitate decision making. The forecasting, different from estimation, is from actual data determine a future condition, such as yields, phenology duration and quality of agricultural products. This work aimed to develop and adjust agrometeorological models to forecast yield and coffee natural drink quality for State of Minas Gerais. We forecasted yields as functions of monthly meteorological variables for three locations in the southern region of the state of Minas Gerais, i.e. Guaxupé, Monte Santo de Minas, and São Sebastião do Paraíso, and for three locations in the Cerrado Mineiro region of Minas Gerais, i.e. Coromandel, Serra do Salitre and Tiros. We used multiple linear regressions selecting independent variables (IV) trying to maximize the anticipation of the yield forecasting for each locality. The IV were the monthly weather elements from 1997 to 2014. The elements used were the air temperature (°C) rainfall (mm), the potential evapotranspiration (mm), the water storage in the soil (mm), the water deficit (mm) and the water surplus (mm). The selection of IV was made testing all possible combinations (APC), avoiding multicollinearity. The models were built with multiple linear regression to maximize the anticipation of forecast. The better models were classified seeking to minimize the mean absolute percentage error (MAPE), determination coefficient (adjusted R²) near 1 and selecting model only with significance at 0.01. The agrometeorological models were accurate for all regions and the lower anticipation period were six and five months for the locations of Guaxupé and Coromandel, respectively. Ranges of monthly values for each meteorological element were established indicating that to reach a high quality of drink at Cerrado it is necessary higher temperatures, lower precipitations, lower storages and higher water deficits when compared to the South of Minas Gerais. (AU) | |
| FAPESP's process: | 14/05025-4 - Agrometeorological models for forecasting yield and natural coffee drink quality |
| Grantee: | Lucas Eduardo de Oliveira Aparecido |
| Support Opportunities: | Scholarships in Brazil - Master |