versión On-line ISSN 1678-4499
SANTOS, Marco Antonio dos y CAMARGO, Marcelo Bento Paes de. Calibration of an agrometeorological model for predicting coffee (Coffea arabica L.) productivity in Sao Paulo State, Brazil. Bragantia [online]. 2006, vol.65, n.1, pp. 173-183. ISSN 1678-4499. http://dx.doi.org/10.1590/S0006-87052006000100022.
Agrometeorological models make possible to assess the quantitative influence of climatic variables, such as air temperature and soil water balance on the coffee development and grain production. An agrometeorological model (CAMARGO et al., 2003) that monitor and assess agrometeorological impact on coffee yields just before the beginning of the maturation growth stage was modified and calibrated. Grain yield were collected from adult coffee plantations at four different regions of the State of Sao Paulo, Brazil. The modified agrometeorological model is based in two parts: first, the model estimates the beginning of the floral induction based on accumulated growing degree days, and a critical rainfall depth. The second part is based on penalization of the potential crop grain yield according to the previous yield and the water stress ratio (ETr/ETp), derived by a 10-day soil water balance during different growth stages. These ratios were weighted by derivation of crop phase yield-response sensitivity coefficients (Ky values), in a multiplicative type model. Also, the model considers penalization for minimum and maximum air temperature. An analysis of the sensitivity coefficients values shows that this model gives higher weight to the water relations during flowering and coffee bean formation phases. This period generally occurs between October and January and it will determine the production of the coffee crop. The statistical analysis for actual and estimated coffee grain yield presented a good linear relationship, "R" between 0.76 and 0.93, "d" index of agreement between 0.73 and 0.90, "C" index of performance between 0.60 and 0.84, and "Ea" unsystematic error between 144 and 558 kg ha-1. The values of "Es" systematic error were relatively low, between 324 and 762 kg ha-1, showing a little tendency of the model to overestimate the predicting coffee grain productivity. The results support the overall conclusion that the proposed model shows a good capacity to estimate coffee grain productivity and it is a promising tool for monitoring climatic impacts on coffee grain yields.
Palabras llave : coffee; productivity; phenology; modelling; sensitivity coefficients; water deficit; air temperature.