ABSTRACT

Acidic mine waste is one of the most common land degradation processes and is a severe environmental problem. The objective of this study is to investigate the potential to predict heavy metal pollution (lead content) in soils with Aster data. The methodology consists of determining the chemical and physical composition, moving through fine resolution waste laboratory spectra measurements and the development of Partial Least Squares Regression (PLSR) models to link these results to Aster imagery. Two Aster scenes were selected in years 2007. Both have been pre-processed and classified by the method of the Spectral Angle Mapper (SAM). The spectra collected under laboratory conditions using spectrophotometer were stored in a spectral library, which was then re-sampled to Aster and, together with the derived spectra from the Aster scenes, was input into the PLSR model collectively with the determined concentrations. Results of the study show that a relationship exists between reflectance spectra and geochemically measured pollutants. Outputs confirm these relationships and also indicate that spectral parameters that were derived from spectral features, reflectance values- in the 400-2500nm range, spectral features re-sampled in Aster band and spectral feature Aster band can directly predict pollutants concentration. The libraries derived from the scenes were also tested against the measured lead content. A relationship was highlighted between lead and the spectral zones according to the following: from 550nm to 584nm, from 1190 to 1200 and from 1220nm to 1256nm. PLSR coefficients were plotted against each other resulting in good predictions and strong relationships established between spectra and lead content (R2>0.7 and RMSE=0,287). The results shown for the relationships between spectral parameters and heavy metal (Pb) concentration indicate that Fe oxides have an important role in the prediction of heavy metals in soil using reflectance spectra.

KEYWORDS

Mine waste, Lead (Pb), Spectroscopy NIR, Aster, PLSR