Produtividade de café arábica estimada a partir de dados de modelos de circulação global
Carregando...
Data
2017-02-21
Autores
Valeriano, Taynara Tuany Borges [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O Brasil é o maior produtor da segunda commodity de maior valor no mercado, o café. O conhecimento de técnicas eficazes de estimativa de produtividade é de grande importância para o mercado cafeeiro, possibilitando melhor planejamento e tornando a atividade mais sustentável. Uma forma eficaz de se estimar a produtividade é através da modelagem agrometeorológica, que quantifica a influência do clima nos cultivos agrícolas. Entretanto, dados climáticos são necessários, e estes em sua maioria são provenientes de estações de superfície. Uma forma de inovar essa técnica é a utilização de outra fonte de dados climáticos, como os dados em grid (GD) que são resultado da combinação de diversas fontes, como sondas oceânicas, sensoriamento remoto, entre outros. O presente trabalho teve como objetivo estimar a produtividade de Coffea arabica a partir do modelo proposto por Santos e Camargo (2006) (SC) com dados meteorológicos provenientes dos GD dos sistemas ECMWF e NASA para regiões cafeeiras de Minas Gerais e São Paulo. Em um primeiro momento foram comparados dados de temperatura do ar e precipitação obtidos pelo ECMWF e NASA aos dados de estações meteorológicas de superfície, com o objetivo de verificação da acurácia dos GDs. No segundo momento foi proposta uma calibração no modelo de estimativa de produtividade de Santos e Camargo (2006), para a utilização dos GDs. Para temperatura, os dados do ECMWF e NASA foram precisos e acurados, com valores mínimos de RMSE iguais a 0,37 e 0,50 °C, e d de Willmott iguais a 0,86 e 0,53, respectivamente. Para precipitação os valores mínimos de precisão e acurácia foram inferiores, com RMSE iguais a 2,15 e 5,33 °C, e d de Willmott iguais a 0,79 e 0,59, respectivamente. Quando os dados GD foram aplicados no modelo SC para estimação de produtividade de café de forma geral o ECMWF foi superior a NASA. A calibração dos coeficientes do modelo SC para dados do ECMWF e NASA foi eficiente, pois houve diminuição do erro percentual absoluto médio (MAPE), raiz do erro médio quadrático sistemático (RMSEs) e raiz do erro médio quadrático não-sistemático (RMSEn) em relação aos modelos SCs que usaram dados de superfície. Estes resultados comprovam que os modelos SCs foram, precisos, acurados e apresentaram baixa tendência, portanto a utilização de dados meteorológicos de GD nos modelos SCs foi possível e recomenda-se a utilização dos mesmos para as regiões cafeeiras de Minas Gerais e São Paulo.
Brazil is the largest producer of the second most valuable commodity on the market, coffee. The knowledge of effective techniques of yield estimation is importance for the coffee market, enabling better planning and making the activity more sustainable. An effective way of yield estimating is through agrometeorological modeling, which quantifies the influence of climate on agricultural crops. However, weather data is needed, and these mostly come from surface stations. One way to innovate this technique is to use another source of climate data, such as Gridded data (GD), which are the result of a combination of diverse sources such as oceanic probes, remote sensing, among others. This work aimed to propose the use of the model of yield estimation proposed by Santos and Camargo (2006) with meteorological data from the GCMs, ECMWF and NASA for coffee regions of Minas Gerais and São Paulo. At first, temperature and precipitation data obtained through the ECMWF and NASA were compared and verified with data from surface meteorological stations. In the second moment, a calibration was proposed in Santos and Camargo (2006), for the use of GDs (ECMWF and NASA). For temperature, ECMWF and NASA data were accurate and accurate, with minimum RMSE values of 0.37 and 0.50 ° C, and Willmott's d values of 0.86 and 0.53, respectively. For precipitation, the minimum values of precision and accuracy were lower, with RMSE equal to 2.15 and 5.33 ° C, and d of Willmott equal to 0.79 and 0.59, respectively. When the GD data were applied in the SC model for coffee yield estimation the ECMWF was superior to NASA. The calibration of the SC model coefficients for ECMWF and NASA data was efficient, since there was a decrease in the mean absolute percentage error (MAPE), mean root mean square root (RMSEs) and root mean square root mean square error (RMSEn) in SCs that used surface data.These results confirm that the models were accurate, accurate and presented a low tendency, so the use of GD meteorological data in agrometeorological models was satisfactory and it is recommended to use them for the coffee regions of Minas Gerais and São Paulo.
Brazil is the largest producer of the second most valuable commodity on the market, coffee. The knowledge of effective techniques of yield estimation is importance for the coffee market, enabling better planning and making the activity more sustainable. An effective way of yield estimating is through agrometeorological modeling, which quantifies the influence of climate on agricultural crops. However, weather data is needed, and these mostly come from surface stations. One way to innovate this technique is to use another source of climate data, such as Gridded data (GD), which are the result of a combination of diverse sources such as oceanic probes, remote sensing, among others. This work aimed to propose the use of the model of yield estimation proposed by Santos and Camargo (2006) with meteorological data from the GCMs, ECMWF and NASA for coffee regions of Minas Gerais and São Paulo. At first, temperature and precipitation data obtained through the ECMWF and NASA were compared and verified with data from surface meteorological stations. In the second moment, a calibration was proposed in Santos and Camargo (2006), for the use of GDs (ECMWF and NASA). For temperature, ECMWF and NASA data were accurate and accurate, with minimum RMSE values of 0.37 and 0.50 ° C, and Willmott's d values of 0.86 and 0.53, respectively. For precipitation, the minimum values of precision and accuracy were lower, with RMSE equal to 2.15 and 5.33 ° C, and d of Willmott equal to 0.79 and 0.59, respectively. When the GD data were applied in the SC model for coffee yield estimation the ECMWF was superior to NASA. The calibration of the SC model coefficients for ECMWF and NASA data was efficient, since there was a decrease in the mean absolute percentage error (MAPE), mean root mean square root (RMSEs) and root mean square root mean square error (RMSEn) in SCs that used surface data.These results confirm that the models were accurate, accurate and presented a low tendency, so the use of GD meteorological data in agrometeorological models was satisfactory and it is recommended to use them for the coffee regions of Minas Gerais and São Paulo.
Descrição
Palavras-chave
Modelagem, Coffea arabica, Dados em GRID, Big data, Modeling, Coffea arabica, Grid data