Modelling and mapping some soil surface properties of Central Kelkit Basin in Turkey by using Landsat-7 ETM+ images

Abstract

Using remote sensing and geographical information system (GIS) tools, we investigated relationships among Landsat-7 Enhanced Thematic Mapper Plus (Landsat-7 ETM+) satellite bands and some soil surface variables of Central Kelkit Basin in Turkey. We collected 164 geo-referenced surface soil samples (0–20 cm) from field studies in 2008 and analysed the data to determine soil variables such as calcium carbonate (CaCO₃), pH, electrical conductivity (EC), organic matter (OM), nitrogen (N), phosphorus (P), exchangeable potassium (EK), and texture (clay, silt, and sand). Utilizing geographic references and determined soil variables, we established a point database in GIS to overlay and extract corresponding digital number (DN) values of Landsat-7 ETM+ bands. We studied possible relationships between the soil variables and the bands' DN values by using correlation (Pearson) analysis. We found that four soil variables including pH, OM, CaCO₃, and N have significant correlations with band 5 (short wave infrared) DN values. We conducted linear regression analysis to develop models for these four variables. Residual statistics and plots showed that the developed models were robust. After running the models in GIS, we mapped pH, CaCO₃, OM, and N contents of the basin.

Acknowledgements

The authors thank the late Dr Mustafa Kilic (1950–2011) for his support in field studies, Michigan State University for sharing their satellite image archives on the Web and Scientific Research Projects Commission of Gaziosmanpaşa University for funding this research (Project No.: BAP 2008/27).