A comparison between various two-variable regression methods in order to estimate soil components using satellite data Case study: Abarkooh playa, Yazd

Abstract

The study of relationship between various soil parameters and satellite data is an effective step in the identification and separation of desert facies. To do so, this study aim to study two-variable regression methods based on different relationships between the various soil components data and ASTER satellite data in order to detect Abarkoo playa facies. To achieve this purpose, at first, 30 topsoil samples were collected from the study area and analyzed in laboratory. Various pedological components (Anion, Cations, Soil moisture, Texture and PH) were also measured. After performing the necessary processing on the satellite images, the value of pixels in each band were extracted by overlaying ground points over satellite image. In the next step, the correlation between satellite data and laboratory values were evaluated by using various two-variable regression methods. The accuracy of the models was assessed using Relative Error, Root Mean Square Error, and Correlation Coefficient of Efficiency. Results indicated that the minimum correlation coefficient is 45%, the maximum relative error of estimation and confirmed are respectively 247.4 and 2489.7 percent, root mean square error is low and the minimum Coefficient of Efficiency is 19 percent. Furthermore, the results of this study showed that there is no significant relationship between PH and soil moisture and satellite data in the study area.

Keywords

• two-variable regression
• kavir facies
• satellite data
• components of soil
• aster
• abarkoo playa