APPLICATION OF POLYNOMIAL REGRESSION IN THE STATISTICAL EVALUATION OFSOIL RESISTANCE

Soil resistance, which indicates the effect exerted by the soil on the cultivating implement, is a characteristic index of the cultivability of the soil. It is also defined as a relative expression of soil comp actedness. In a long-term soil cultivation experiment set up on brown forest soil in Godollo in 1994, the effect of five cultivation techniq ues (direct drilling, disking, ploughing. loosening+disking, loosening +ploughing) on soil resistance was examined. The soil resistance was d etermined in four replications for each cultivation technique, at 15 d epths between 0 and 70 cm in each replication. The statistical analysi s of the results was primarily aimed at describing the compaction proc ess for each cultivation mode and at determining and comparing the dep th of maximum soil compactedness. The analysis consisted of the follow ing steps: 1. Graphic plotting of the measured data to choose the thre e replications most characteristic of the cultivation mode. 2. Reducti on of the random effects occurring in each data series by calculating a moving average. 3. Fitting of four types of single-variant polynomia l regression to describe the compactedness of the soils: second order polynomial regression (quadratic effect curve), third order polynomial regression. third order polynomial regression without a linear compon ent, third order polynomial regression without a first order component or a constant. The best fit was given by this last function. The fitt ing of all four types of regression was carried out for three differen t soil depths (30, 45 and 60 cm). 4. Validation of the polynomial regr ession model and the results obtained. 5. Estimation of the maximum de pth and value of soil resistance for each cultivation mode, based on t he polynomial regression best describing the compaction process.