Evaluating some penetration resistance models in some Khozestan sugarcane agro-industries

Functions that accurately relate penetration resistance to soil properties are important tools for assessing the spatial variation of soil mechanical resistance when the temporal and spatial variability in these parameters cannot be readily measured. The importance of this characteristic has encouraged the soil researchers to model this process for quantitative applications. In this study, the performance of twelve penetration resistance models were evaluated for two depths. i.e. 0-40 and 40-80 cm. For this purpose, the penetration data were obtained using of digital penetrometer from 150 points of different Khuzestan sugarcane agro-industry. The parameters of these models were then determined using the least square optimization method. In order to evaluate the accuracy of the models, the mean error (ME), root mean square error (RMSE), mean absolute of mean error (MAME) and Pearson correlation coefficient (r) statistics were calculated. The results of comparison models performance revealed that the whole proposed regression models based on moisture content and bulk density had lower error and higher r in estimating the penetration resistance. Furthermore, the models proposed by To and Kay based on pore water suction and Dexter et al. based on the slope of the water retention curve at its inflection point and effective stress have highest error and lowest r in estimating the penetration resistance. Generally, all of the evaluated models were systematically over-estimator models.