An Assessment of Pedotransfer Function Performance for the Estimation of Spatial Variability of Key Soil Hydraulic Properties. Issue 9 (7th September 2017)

• Record Type: Journal Article
• Authors: da Silva, Alessandra Calegari; Armindo, Robson André; Brito, Alexsandro dos Santos; Schaap, Marcel G.
• Is Part Of: Vadose zone journal. Volume 16:Issue 9(2017)
• Page Start: 1
• Page End: 10
• Publication Date: 2017-09-07
• Subjects: Soil science -- Periodicals; Zone of aeration -- Periodicals; Groundwater flow -- Periodicals; Groundwater flow; Zone of aeration; Periodicals; Electronic journals; 631.4
• DOI: 10.2136/vzj2016.12.0139
• Languages: English
• ISSNs: 1539-1663

Abstract

Abstract : Core Ideas: PTFs have been widely applied in environmental and hydrological watershed studies. PTFs can simulate the spatial variability of hydraulic functions. The simulation of hydraulic maps using PTF data should be performed with care. The reliable applications of soil water dynamics and its spatial variability are very important in hydrological and environmental studies. However, determination of some hydraulic properties requires time and sometimes high costs. Pedotransfer functions (PTFs) have been applied as an alternative for the prediction of soil hydraulic properties by working with some easily measurable physical properties. This study examined the performance of the Splintex and Rosetta PTFs for estimating the spatial variability of some soil hydraulic properties such as available water capacity (AWC), field capacity (θfc ), permanent wilting point (θpwp ), specific water capacity [ C (θ)], conductivity [ K (θ)], and diffusivity [ D (θ)] in a sandy soil from southeastern Brazil. Based on the log‐likelihood, Akaike, and Bayesian information criteria, semivariogram models were fitted and used to krige the assessed variable maps. The results showed the feasibility of using these PTFs to estimate the spatial variability in soil hydraulic properties and functions, which would greatly benefit vadose zone flow and transport modeling. Geostatistical analyses presented similarities in kriged maps of both observed K (θ) and C (θ) data with kriged maps by Rosetta estimates, as well as the kriged map of the observed AWC data with the kriged map by Splintex estimates sometimes revealing the same semivariogram data pattern. Predicted maps of θfc, θpwp, and D (θ) were, however, different from kriged maps of the observed data, showing that this approach should be used carefully. The use of the range parameter from the semivariogram models was efficient to determine the optimal sample density for the studied area.