Анатолий Пуховский » Публикация

Аннотация Abstract—The type of mathematical model of the production function is very important for an adequate pre diction of optimal fertilizer rates and their effect on yield. In this case, easily calibrated regression models can’t be used for extrapolation, and the models proposed by Mitscherlich, Spillman, Baule, and von Bogus lawski are difficult to calibrate even in the onedimensional case. An explanation of the good approximating capacity of logarithmic transformation of the argument of the linearizing ascending branch of the production function is found on the basis of model investigations. An algorithm for simplified calibration of the von Boguslawski mathematical model is developed: (1) transform fertilizer rates into a logarithmic form after introducing a correction for the initial content of the nutrient element in soil; (2) estimate the maximum yield under experimental conditions; convert the yield data into a relative form; (3) discriminate the data based on the interval permissible for approximation; (4) find the parameters of the linear model by the leastsquares method and on their basis calculate the parameters of the von Boguslawski model, thereby correcting the form of the approximation and obtaining the possibility of its extrapolation.