Modeling of available soil phosphorus (ASP) using multi-objective group method of data handling

Publication Type:
Journal Article
Modeling Earth Systems and Environment, 2016, 2, (3), pp. 157
Issue Date:
Filename Description Size
Keshavarzi2016_Article_ModelingOfAvailableSoilPhospho.pdfPublished version1.25 MB
Adobe PDF
Full metadata record
The accurate estimation of available soil phosphorus (ASP) is required for the effective management of environmental resources and improving agricultural productivity. Direct measurement of ASP at large scales is time-consuming and expensive. Therefore, many empirical models or pedotransfer functions (PTFs) have been developed to obtain ASP. This study introduces a new model that estimates ASP from easily measurable soil properties by using the multi-objective group method of data handling (mGMDH) approach. Organic carbon, clay content, calcium carbonate equivalent, pH and ASP were obtained from 100 soil samples, which were collected by the stratified random sampling strategy in Mashhad Plain, Khorasan-e-Razavi Province, Northeast Iran. The suggested model for the prediction of ASP generated satisfactory results in terms of the statistical performance criteria. The coefficient of determination (R2) and mean square error (MSE) values of the mGMDH model in training step were 0.606 and 4.9671 mg kg−1, respectively, whereas those were 0.504 and 5.6685 mg kg−1, in the testing step. It was concluded that only 50 % of variations of ASP could be described by selecting easy-to-measure soil variables and owing to the key role of phosphorus in soil fertility and precision agriculture, the other related factors such as terrain attributes and fertilizer management should also be considered as they influence the distribution of the soil phosphorus.
Please use this identifier to cite or link to this item: