Laboratory evaluation of the PLSR method to estimate Atterberg limits of soil by field spectroscopy

Generally in developing countries such as Iran, the studies in soil mechanics are rarely done or even not conducted due to high cost and time of research and conventional laboratory methods, so the lack of information in this field can damage the construction. Today new approaches such as field spectroscopy are being developed in order to dissolve many of these issues, which is a
non-destructive, fast and inexpensive method. Spectrometry is a non-destructive, fast and inexpensive method that has not been applied in estimation of mechanical properties of forest soils in Iran yet. In this study, we estimated the Atterberg limits of soil (Liquid limit, Plastic limit, Plasticity index, and Shrinkage limit) by field spectroscopy. For this purpose, 45 soil samples were collected from Kheyrood forest (NAMKHANE district) to do Atterberg Limits and VNIR. Partial Least Square Regression (PLSR) was used to predict the Atterberg limits in both the raw spectra and the first derivative of the spectrum. The results showed that PLSR modeling for the parameters of Liquid Limit and Plasticity Limit in the raw and first derivative is very suitable and for Plastic Limit and Shrinkage Limit is weak and not acceptable. Therefore spectroscopy can be used as a fast, inexpensively and accurately acceptable method for predicting the Atterberg limits.