Soil bulk density (BD), which can be measured by several labor-intensive procedures, is frequently missing from soil databases. However, it is an essential parameter in many calculations and models, and pedotransfer functions (PTFs) can be developed to estimate it. In this article, the predictive accuracy of 19 published PTFs was evaluated using soil data sets from China. In addition, exploratory stepwise regression models were proposed and validated. The data used in model development were legacy data from various sources and were divided randomly into two sets: a training set for model development with 75% of the data and a validation set for model validation with 25% of the data. The results show that existing models, developed by Alexander (1980) (P1), Manrique and Jones (1991) (P7), and Périé and Ouimet (2008) (N6), respectively, produced relatively accurate predictions. However, the first two models were inappropriate for soils containing a large amount of soil organic carbon. The exploratory model (Model 1) indicated that soil organic matter, organic matter0.5, total nitrogen, and clay were the four most important factors in BD prediction. The exploratory model and its simplified version (Model 3) had higher prediction accuracies than previously published PTFs. The results show that parameters tailored to the current data improved prediction accuracy for the nonlinear model (Model 2). Compared with the exploratory model (Model 1), its simplified version and the nonlinear model, with only one variable, had good prediction accuracies as demonstrated by validation.
1State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China.
2Graduate University of Chinese Academy of Sciences, Beijing, China.
Address for correspondence: Dr. Gan-Lin Zhang, State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, 71 E. Beijing Road, Nanjing 210008, China. E-mail: email@example.com
Received January 16, 2011.
Accepted for publication November 1, 2011.
Financial Disclosures/Conflicts of Interest: This research was supported by the Natural Science Foundation of China (grant 40625001), the Natural Science Foundation of Jiangsu Province (BK2008058) and the Knowledge Innovation Program of the Chinese Academy of Sciences (grant KZCX2-YW-409).