Ali Akbar Noroozi; Mahdi Homaee; Abbas Farshad
Abstract
Soil salinity is a limiting factor for plant growth and a serious cause of land degradation. Field sampling and statistical analysis for estimating soil salinity is expensive and time consuming. Estimating soil salinity by spatial statistical models and Geographic Information System (GIS) is recommended, ...
Read More
Soil salinity is a limiting factor for plant growth and a serious cause of land degradation. Field sampling and statistical analysis for estimating soil salinity is expensive and time consuming. Estimating soil salinity by spatial statistical models and Geographic Information System (GIS) is recommended, because it saves labor and time. This study was conducted to evaluate the performance of spatial statistics with ordinary least square (OLS) incorporation with LANDSAT data to predict soil salinity. The electrical conductivity (EC) of 236 soil samples were collected from Garmsar plain at south east Tehran, Iran and were measured and correlated to 27 variables derived from LANDSAT images, including vegetation indices, salinity indices, bands 1 to 7, principal component analysis and tasseled cap indices. Using factor analysis and similarity index, these variables were divided into three components. Furthermore, two models for soil salinity estimation were derived, using the best correlation correlation coefficient (0.58 and 0.60) method. Simultaneously, soil salinity map was produced in ArcGIS by spatial statistics model ordinary least square (OLS) followed by derivation of the error map, calculated using Moran's index. The error map indicated that the spatial statistics models are superior to classic statistics methods, due to high accuracy in estimation and the fact that it doesn't require exchange information between different software programs
