Retrieving Soil Hydraulic Properties by Diffuse Spectral Reflectance Data in Vis-NIR-SWIR Range

Friday, 19 December 2014
Ebrahim Babaeian1, Mehdi Homaee1, Harry Vereecken2, Carsten Montzka3, Ali Akbar Norouzi1 and Martinus Van Genuchten4, (1)Tarbiat Modares University, Tehran, Iran, (2)Forschungszentrum Julich GmbH, Julich, Germany, (3)Forschungszentrum Jülich, Jülich, Germany, (4)Federal Univ. of Rio de Janeiro, Mechanical Engineering, COPPE/LTTC, Rio de Janeiro, Brazil
Information about the soil water characteristics is necessary for modeling water flow and solute transport processes in vadose zone. Soil spectroscopy in the visible, near-infrared and shortwave infrared (Vis-NIR-SWIR) range has been widely used as a rapid, cost-effective and non-destructive technique to predict basic soil properties. In this paper we used three different approaches to retrieve soil hydraulic parameters from spectral data in the visible, near-infrared and shortwave-infrared (Vis-NIR-SWIR) region and basic soil properties. Using stepwise multiple linear statistics coupled with bootstrapping, we derived and validated three types of point and parametric transfer functions: i) spectral transfer functions (STFs), ii) pedotransfer functions (PTFs) and iii) spectral pedotransfer functions (SPTFs) which respectively used spectral data, basic soil properties and spectral based basic soil predictions as their inputs. We further evaluated a direct fit of the van Genuchten (VG) and Brooks-Corey (BC) retention models to the predicted water contents obtained with each approach. According to the results, soil water contents, the VG and BC parameters as well as basic soil properties showed significant (p<0.01) correlation with spectral reflectance values, especially for the SWIR region. The STFs performed slightly better than the PTFs in terms of R2 and RMSE in estimating water contents in the mid and dry parts of retention curve. In the wet range, PTFs were found to perform better than the other two approaches. Compared to the STFs, however, better water content estimates were obtained using the SPTFs in the wet range. The parametric STFs and SPTFs of both the VG and BC models developed from spectral data performed slightly better than parametric PTFs for the retention curve. The best predictions were obtained with a direct fit of the retention models to soil water contents estimated with point transfer functions. Our findings suggest that spectral information, as a promising approach, may be used to accurately predict soil water contents, and indirectly the water retention curve. Using spectral data as an input of PTFs provides an effective way of including this temporal dynamic soil property in soil water retention predictions.