We conducted a remote sensing analysis to discern features and patterns of areas affected by salt. Maximum likelihood
classification (MLC), Support Vector Machine (SVM), and Minimum Distance (MD) with four classes (slightly,
moderately, high and extreme saline soil) are applied to classify the salt affected areas. 102 samples, collected from the
investigated region, are used as input data set .
Soil properties, land use and ground water table are selected as the main parameters affecting soil salinity. These
parameters are used to understand the spatial distribution of the different classes of soil salinity. Our approach was
applied on hyperspectral data from the EO-1 Mission. The present study highlighted that gypsum soil is obviously fitting
with class of extreme and high saline soil. Thus, high content of gypsum in soil is the most important parameter
controlling the soil salinity in this region. Moreover, water logging is lightly affecting the soil salinity through the rising
of the water table level by sea water seeping; especially in the irrigation areas located no more than 5 km from the coast
line.
Computed accuracy from the classification gave different but encouraging accuracy results varying between 46% and
75%. SVM is showing the best performance in extracting patterns and features of soil salinity classes (kappa coefficient
of 63% and overall accuracy of 75%). Furthermore, this work reveals the high potential of hyperspectral data in
discerning areas that are highly and extremely affected by salinity.
Soluble salts in soils seriously compromise agricultural productivity around the world. Arid and semi-arid regions are most prone to salinization. Careful monitoring and surveying of salt-affected soils is needed to
ensure sustainable development in such regions. Remote sensing techniques are being increasingly applied to
investigate this phenomenon. Our approach is to map low and moderately salt-affected soils in northeast
Brazil through the combination of remote sensing data and geochemical ground-based measurements.
Spectral properties, salinity, vegetation and brightness indices were used to extract salinization features and
patterns from the Brazilian soils. MODIS Terra data were selected to cover the 1.7 million km2 area and the
images were taken during the summer 2008 sampling campaign. The electrical conductivity (EC) from 112
sites was determined (1:5 soil/water suspension method) to test the capability of each indicator to identify
salt-affected areas based on correlations between indicators and electrical conductivity (ground truth).
Eighteen indices emerged from the MODIS Terra images. A moderate correlation was found between
electrical conductivity and the spectral indices. Salinity emerged as the most significant index. Spectral
properties were used to define soil classes based on their degree of salinization.
Near infrared (NIR) region from the electromagnetic spectrum showed high potential to separate different
categories of salt-affected soil from MODIS multispectral data. A low correlation between vegetation indices
and electrical conductivity indicates that these indices are inadequate when trying to discern features and
patterns of salt affected areas on a large scale
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.