Modelos de determinação não-destrutiva da área foliar de videiras viníferas

Abstract

The growing of grapevines has gained importance in new regions, which should be conducted modeling studies of the development to know yield performance of the varieties to be cultived. One of the most important variables for modeling the development is the leaf area, responsible for innumerable interactions between the plant and the environment. Because this contest the aim with this work was generate and test estimate models of the leaf area of grapevine 'Tannat' and 'Merlot' in function of the linear dimensions of the leaf. Was collected vine branches having leaves of all ages, in laboratory, the leaves were measured in the dimensions of length (L) and width (W), the area of the leaf surface (LA) was obtained from reading in an optical integrator leaf area, leaves smaller than 3 cm in length were discarded because they were considered photosynthetically inactive. Data collected from odd and even leaves were stored in separate databases, using the measurements obtained in the leaves odd to fabricate mathematical models of data adjustment and even to test the models. Were created linear, quadratic, exponential, logarithmic and potential models, as a function of L, W and the product L W (LW), being tested the models generated by determination coefficient greater than 0.90. To evaluate the models tested, was calculated the root of the mean square error, subdivided in mean square error systematic and non-systematic, the systematic error was subdivided in mean square error additive, mean square error proportional and mean square error interdependent, BIAS index (%), correlation coefficients of Pearson, concordance, confidence coefficient, efficiency, mean absolute error, root mean square error normalized and coefficient of residual mass. The models were classified according to the proportional value, attributed to their performance in the statistic performance in the statistic performed. Were obtained seven models to 'Tannat' and three models to 'Merlot' with determination coefficient higher than 0.90. When tested the models generated using only one dimension of the leaf did not show satisfactory estimate of the real LA. It is possible to estimate the leaf area of grapevines 'Tannat' and 'Merlot' based on the decision of the linear dimensions of the leaves, using mathematical models, it is suggested to use the model (LA = 0,6734 LW^1,1034) to estimate the leaf area of 'Tannat' and model (LA = 1,2056 + 1,0025 LW) to estimate the leaf area of 'Merlot'.