Modelagem agrometeorológica para previsão de produtividade de açaizeiro
Carregando...
Data
2017-07-31
Autores
Moraes, José Reinaldo da Silva Cabral de
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Modelos agrometeorológicos para a previsão da produtividade permitem a avaliação quantitativa de elementos meteorológicos com as variáveis fenológicas e produtividade, auxiliando no planejamento de safras agrícolas para otimizar estratégias e tomada de decisão. Na região amazônica, mais precisamente no estado do Pará, concentram-se as maiores produções de açaí no Brasil, entretanto a sazonalidade na produção, influenciada principalmente pelo clima que altera fortemente o preço do produto, tornando-se preocupante para toda cadeia produtiva. Portanto analisar o desempenho de modelos de circulação geral (GSM), que possam representar as condições de tempo e clima na região é de grande importância para a sustentabilidade da produção. Objetivou-se avaliar os dados de temperatura do ar (T) e precipitação (P) provenientes do modelo do European Center for Medium-Range Weather Forecast (ECMWF) e desenvolver modelos agrometeorológicos para a previsão da produtividade em cultivos de açaizeiro em Tomé Açu, Pará. Dados de T e P de estações meteorológicas de superfícies foram estimados por regressão linear usando dados em grid do (ECMWF), correspondendo ao período de 1990 a 2014, separando-se em escala decendial e mensal. Com base nos dados estimados foram testados e ajustados modelos de regressão linear múltipla para prognosticar a produtividade em dois tipos de cultivo de açaizeiro. Buscou-se maximizar a data para antecipação da previsão. A variável dependente foi à produtividade da cultura, enquanto que, as variáveis independentes foram os elementos meteorológicos decendiais e mensais, para cultivos irrigado (CI) e não irrigado (CNI). O ECMWF tende a superestimar precipitação pluvial na estação seca e subestimá-lo na estação chuvosa. A região noroeste da Amazônia é caracterizada como a área de maior oferta de P, mas com menores desempenhos do modelo com R2 inferior a 0,18. O modelo ECMWF apresentou alto desempenho (R2> 0,60) ao estimar P em escala mensal e desempenho médio (R2<0,60) ao estimar T em período mensal e decendial. As maiores precisões do ECMWF foram nas áreas antrópicas no sudeste e nordeste da região amazônica, precisamente onde localizam-se os pólos de produção de açaí. A fase de folha madura e emissão de espata, para CI e CNI foram as mais sensíveis às condições climáticas. As maiores produções de frutos por planta ocorreram nos meses de novembro e dezembro, nos dois sistemas de cultivo. As observações apontaram produtividades que variaram entre 16.500 kg ha-1 em CI e 7.500 kg ha-1 em CNI. Os modelos agrometeorológicos foram acurados com valores máximos de MAPE de 0,01 no CI e 1,12 no CNI. Foi possível a previsão da produtividade em média de 6 a 9 meses antes da colheita, em abril, maio, novembro e dezembro no CI e em janeiro, maio, junho, agosto, setembro e novembro para CNI.
Agrometeorological models for yield prediction allow the quantitative evaluation of the meteorological elements with the phenological variables and yield, aiding in the planning of agricultural crops and in the optimization of decision making. In the Amazon region, more precisely in the Pará State, the largest açai production is found in Brazil. However, seasonality in production, influenced mainly by the climate, is strongly affecting the price of the product, causing concern for the entire production chain. Therefore, evaluating the performance of general circulation models (GSM), which may represent the weather and climate conditions in the region, is of great importance for the sustainability of production. The objective of this study was to evaluate the air temperature (T) and precipitation (P) data from the European Center for Medium-Range Weather Forecast (ECMWF) model and to develop agrometeorological models for yield prediction in açai crop in Tomé Açu, Pará. T and P of meteorological stations of surfaces were estimated by linear regression using grid data of the ECMWF, from the period 1990 to 2014, separating in 10-day period and monthly scale. Based on the estimated data, multiple linear regression models were tested and adjusted to predict yield in two types of açai culture. We seek to maximize the date for anticipation of the forecast. The dependent variable was crop yield, while the independent variables were meteorological elements for irrigated system (IRRS) and rainfed system or unirrigated (RAINF). The ECMWF tends to overestimate rainfall in the dry season and to underestimate it in the rainy season. The northwest region of Amazonia is characterized as the area of highest rainfall supply (P), but with lower performances of the model with R2 less than 0.18. The ECMWF model presented a high performance (R2> 0.60) when estimating P on monthly scale and mean performance (R2<0.60) when estimating T in monthly and 10-day period. The most important ECMWF accuracies were in the anthropic areas in the southeast and northeast of the Amazon region, precisely where the açai production poles are located. The phase of leaf maturation and emission of the spathe, for both IRRS and RAINF, were the variables that were most sensitive to climate conditions. The greatest fruit production occurred in November and December in each cultivation system, and production ranged from 7.500 kg ha-1 in the RAINF to 16.500 kg ha-1 in the IRRS. The accuracy of the agrometeorological models, using maximum values for average absolute percent error (MAPE), was 0.01 in the IRRS and 1.12 in the RAINF. Using these models production was predicted approximately 6 to 9 months before the harvest, in April, May, November, and December in the IRRS, and January, May, June, August, September, and November for the RAINF.
Agrometeorological models for yield prediction allow the quantitative evaluation of the meteorological elements with the phenological variables and yield, aiding in the planning of agricultural crops and in the optimization of decision making. In the Amazon region, more precisely in the Pará State, the largest açai production is found in Brazil. However, seasonality in production, influenced mainly by the climate, is strongly affecting the price of the product, causing concern for the entire production chain. Therefore, evaluating the performance of general circulation models (GSM), which may represent the weather and climate conditions in the region, is of great importance for the sustainability of production. The objective of this study was to evaluate the air temperature (T) and precipitation (P) data from the European Center for Medium-Range Weather Forecast (ECMWF) model and to develop agrometeorological models for yield prediction in açai crop in Tomé Açu, Pará. T and P of meteorological stations of surfaces were estimated by linear regression using grid data of the ECMWF, from the period 1990 to 2014, separating in 10-day period and monthly scale. Based on the estimated data, multiple linear regression models were tested and adjusted to predict yield in two types of açai culture. We seek to maximize the date for anticipation of the forecast. The dependent variable was crop yield, while the independent variables were meteorological elements for irrigated system (IRRS) and rainfed system or unirrigated (RAINF). The ECMWF tends to overestimate rainfall in the dry season and to underestimate it in the rainy season. The northwest region of Amazonia is characterized as the area of highest rainfall supply (P), but with lower performances of the model with R2 less than 0.18. The ECMWF model presented a high performance (R2> 0.60) when estimating P on monthly scale and mean performance (R2<0.60) when estimating T in monthly and 10-day period. The most important ECMWF accuracies were in the anthropic areas in the southeast and northeast of the Amazon region, precisely where the açai production poles are located. The phase of leaf maturation and emission of the spathe, for both IRRS and RAINF, were the variables that were most sensitive to climate conditions. The greatest fruit production occurred in November and December in each cultivation system, and production ranged from 7.500 kg ha-1 in the RAINF to 16.500 kg ha-1 in the IRRS. The accuracy of the agrometeorological models, using maximum values for average absolute percent error (MAPE), was 0.01 in the IRRS and 1.12 in the RAINF. Using these models production was predicted approximately 6 to 9 months before the harvest, in April, May, November, and December in the IRRS, and January, May, June, August, September, and November for the RAINF.
Descrição
Palavras-chave
Amazônia, Dados em GRID, Euterpe Oleracea, Modelos de previsão, Amazon, Data in GRID, Prediction models