Date

2013

Author

Soydan, Hilal

Metadata

Show full item record

Item Usage Stats

231
views

302
downloads

Cite This

Being an indispensable economic and energy source, coal mining may be the cause of long term environmental damages like serious water and soil contamination due to inappropriate mining operations. In order to detect and avoid it, mapping the contamination levels related to mining industry has been a challenge for soil scientists to find the most appropriate and time and cost effective method for constructing soil contamination maps. So far, Remote Sensing (RS) and Geographical Information Systems (GIS) have been utilized as auxiliary data for producing more accurate Digital Soil Maps (DSM), especially by methods based on supervised classification techniques. In this study, Worldview 2 stereo imagery and ASTER imagery of an abandoned mining area were utilized for data analyses by forming statistical relationships between reflectance spectra of both imageries, and results of chemical analyses of soil. Similar statistical relationships are examined between the laboratory reflectance spectra and the chemical properties of the soil samples from study area. After employing necessary pre-processing techniques, Worldview 2 and ASTER data spectra were utilized for Multiple Linear Regression (MLR) analysis. Furthermore, after the fieldwork, ASD Inc. Agrispec spectrometer measurements were performed in order to collect laboratory spectra of soil samples as an input of spectral library of the area and the acquired spectra data were used as the input for Partial Least Square Regression (PLSR) analyses for demonstrating the relationship between the constituents of soil and laboratory spectra. The PLSR coefficients were plotted for identifying important wavelengths related to the properties and a comparison between PLSR and MLR analysis results was performed to see the consistency and differences of predictive wavelengths for each soil property. The comparison showed that the results were internally consistent as well as with the published results in the literature to predict specified soil properties. In addition, MLR equations defining the soil properties were applied to the bareland areas of the study area to produce contamination maps whose accuracy assessments were implemented with an independent data set. ASTER, whose bands up to shortwave infrared region were utilized, ended up providing higher accuracies than Worldview-2 for producing soil contamination maps.

Subject Keywords

Coal mines and mining., Coal mines and mining

URI

http://etd.lib.metu.edu.tr/upload/12616304/index.pdf
https://hdl.handle.net/11511/22947

Collections

Graduate School of Natural and Applied Sciences, Thesis