Comparing Pedotransfer Functions to Estimate Soil Bulk Density in Central Chile.

Bulk density (BD) is a fundamental soil property that is typically required for predicting soil processes such as compaction, infiltration, and runoff. In the absence of field measurements, pedotransfer functions (PTFs) have been developed to estimate BD. Because they are built on existing measure data, the accuracy and reliability of the PTFs depends on the data sets used for calibration and validation. Thus, many already available PTFs lack of universal applicability. Therefore, the objective of this study was to test the performance of a series of already existing PTFs and assess the improvements when the functions were calibrated using an independent soil dataset from Central Chile. A total of 1,008 spatially distributed BD data points were used from 243 soil series. The tested PTFs were developed by Adams (1973), Saxton et al. (1986), Bernoux et al. (1998), Tomasella and Hodnett (1998), Kaur et al. (2002), Rawls et al. (2004), Heuscher et al. (2005), Saxton and Rawls (2006), Benites et al. (2007), and Brahim et al. (2012). The results showed that, depending on the PTF, the R² ranged between 0.06 and 0.55 and the RMSPE between 0.24 and 0.46 before calibration. Post-calibration estimates showed an R² between 0.17 and 0.62 and an RMSPE between 0.20 and 0.33. The Heuscher’s relationship, which is the only that use the soil water content at wilting point, showed the highest R² and the lowest RMSPE. In contrast, the Saxton et al. relationship showed the lowest R² and the highest RMSPE, which is mainly explained because it only uses clay and sand contents, variables that in these soils were not highly correlated with BD. This study provides validation of different PTFs and discusses the relevance of using them on Chilean soils.