In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
ENVIRNOMENTAL IMPACT ASSESSMENT USING NEURAL NETWORK MODEL: A
CASE STUDY OF THE JAHANI, KONARSIAH AND KOHE GACH SALT PLUGS, SE
SHIRAZ, IRAN
M. H. Tayebi a ’ *, M. H. Tangestani a , H. Roosta b
d Dept, of Earth sciences, Faculty of sciences, Shiraz University, 71454, Shiraz, Iran- Mhtayebi@shirazu.ac.ir-
T angestani@susc. ac. ir
b Dept, of Civil Engineering, Marvdasht Branch, Islamic Azad University, Marvdasht, Iran -
Hasan. Roosta@gmail. com
KEY WORDS: Salt plug, Environmental impact, MLP neural network, ASTER
ABSTRACT:
This study employs Multi-Layer Perceptron (MLP) to estimate environmental impact of salt plugs using Advanced Spacebome
Thermal Emission and Reflection Radiometer (ASTER). VNIR and SWIR datasets of ASTER were assessed in mapping and
detecting Jahani, Konarsiah, and Kohe Gach salt plugs and the affected areas located at SE Shiraz, Iran. PC color composite and
geological map of the region were used to select training areas. Three datasets including, IARR, PCA and MNF were used as input
to the MLP. The results of each input were compared with the ground truth and the geological map to determine the accuracy and
therefore to select the more appropriate dataset to be input to MLP approach input. The results demonstrated a number of the
polluted sites and the main polluted tributaries that convey the water as well as the salt plug materials into the Firouzabad River. It is
also indicated that the MNF input (with 85% overall accuracy) can obtain a slightly more accurate estimation than the IARR (79%)
and PCA inputs (82%). It is concluded that the result of MNF input to MLP is more applicable to effective environmental impact
assessment and sustainable water resources management at salt plug-affected areas.
1. INTRODUCTION
Salinity caused by natural processes is a major envimomental
hazard and can have hazardous effects on agricultural
production, water quality, ecological health, soil erosion, flood
risk, infrastructure and the society. The effects and damages of
salinity are not stronger than earthquake or landslide
(Mettemicht and Zink, 2003), but it is a major threat in semi-
arid and arid regions such as Iran. The most important impact of
salinity is salinization of fresh rivers, which affects the quality
of water for drinking and irrigation.
More than 150 known salt plugs (Kent, 1970) are exposed at the
south eastern Zagros Folded Belt, southern Iran. These saline
formations are important because: (1) they can potentially trap
the hydrocarbons, (2) for their potential in ore deposition, and
(3) they can provide harmful environmental impacts. Three of
these salt plugs, namely Konarsiah, Jahani and Kohe Gach are
exposed at the SE Shiraz, southern Firouzabad (Fig. 1). These
salt plugs increase the salinity of groundwater, surface water
(especially Firouzabad River), and the adjacent soils by direct
dissolution and transport of soluble salt plug minerals, which
directly influence the economy and ecosystem of the area.
Information on the extent of the salt plug-affected areas is
required for effective environmental planning and sustainable
water resources management. Assessing the spread of salinity
by salt plugs has traditionally been implemented by
geochemical, hydrologic, and geophysical (Zadneek, 2008;
Ghanbarian, 2007; Dehghan, 2008) methods requiring the
collection of numerous samples followed by laboratory
measurements. However, remote sensing can act as an effective
means of detecting environmental pollution and is a useful tool
for acquiring basic information particularly on a regional scale
(Sabins, 1997). The task of identifying salinity largely depends
on the peculiar way salts distribute at the soil surface and within
the soil mantle, and on the capability of the remote sensing
tools to identify salts (Zinck, 2001). Many remote sensing
techniques and datasets have already been used to map salt-
affected areas (Hunt and Salisbury, 1976; Hick and Russell,
1990; Mougenot et al., 1993; Ben-Dor et al., 2002; Mettemicht
and Zink, 2003; Farifteh et al., 2006), but there is lack of a
publication focusing on the application of remote sensing in
mapping and detecting the salt plug environmental impact. An
unpublished work of Tavakkoli (2008), however, used the
ASTER data for enhancing the lithological units of the same
salt plugs.
Artificial neural network (ANN) is an interconnected group of
nodes using mathematical methods to process information. It is
a self adaptive system, which can change its structure based on
the internal or external information (Hu and Weng, 2009).
Among all the techniques, artificial neural networks (ANN)
have been widely used (Ji, 2000, Zhai et al., 2006) due to its
advantages over statistical methods (Bischof et al., 1992) such
as no assumption about the probabilistic models of data, robust
in noisy environments, and the ability to learn complex patterns
(Ji, 2000). Neural networks have been applied in the large
number and wide variety of applications (Liu et al., 2001;
Kavzoglu & Mather, 2003; Verbeke et al., 2004; Chormanski et
al., 2008; Hu and Weng, 2009). The primary aim of this study
was identifying and mapping the salt plugs as well as the salt
plug-affected areas. The second aim is to evaluating the use of
Corresponding author. Dept, of Earth Sciences, Faculty of Sciences, Shiraz University, 71454 Shiraz, Iran. Tel: +9809177173319 ;
fax: +982284572. Mhtayebi@shirazu.ac.ir.
