A study on land use/cover classification with textural analysis using Multi-Temporal JERS-1 (SAR) L- band in arid and semi-arid areas.(A case study in Northeastren Jordan)
Hussam AL- Bilbisi
Ph.D. Student, Center for Environmental
Remote Sensing (CEReS), Chiba University
1-33,Yayoi- cho, Inage- ku, Chiba, 263-8522
Tel: (81) 43 290 3965
Fax:(81) 43 290 3857
E mail: hbilbisi@ceres.cr.chiba- u.ac.jp
Japan
Ryutaro Tateishi
Professor, Center for Environmental
Remote Sensing (CEReS), Chiba University
1-33,Yayoi- cho, Inage- ku, Chiba, 263-8522
Tel: (81) 43 290 3965
Fax:(81) 43 290 3857
E mail: Tateishi@ceres.cr.chiba-u.ac.jp
Japan
Abstract
A subset of each of the of the JERS-1 SAR L-band images acquired in October 1993, and March 1995,
respectively, were used for land use/cover classification purpose in a part of northeastern Jordan area.
Textural feature images were generated using JERS-1 SAR L-band 4-look images. These images had
been used as input for supervised classification purpose. The accuracy of various classified images had
been compared.
This paper describes the methodology and discusses the results for the purpose of textural feature
analysis, and land use/cover classification in arid and semi-arid areas (as for the studied area case)
using SAR L-band data set.
Introduction
Satellite remote sensing is a useful technique to map land use / cover, where this technique offers
unparalleled technique to monitor urban, green and non- green lands changes at regional, global and
local scales from space. Although northeastern Jordan is accounted as an arid to semi-arid region, it is
dominated by thick blanket of surficial deposits and in some cases haze and clouds can determine the
quality of the optical satellite data. However, Synthetic Aperture Radar (SAR) imaging types can provide
clear images in all weather conditions and can indeed help in the land use/cover in arid and semi arid
areas. In general (SAR) images have speckle noise, which causes salt and- pepper noise, which in turn
affect the classification accuracy of the classified images.
Textural feature images have been widely used in classification to improve the ability to discriminate
objects (Takashi et al. 2001). Therefore textural analysis may be a powerful tool for improving the
classification accuracy. Various textural measures have been investigated for estimating the
performance of textural analysis.
In this study, and by using JERS-1 (SAR) L- band, we investigate a method to evaluate the accuracy of
the land use/cover classification using the original, smoothed and textural feature images in arid and
semi-arid areas.
Study Area
The study area (Fig.1) is located in the northeastern of Jordan, and includes a small city named
rewaished. It lies between the latitude 32°15' 22' and 32°50' 17' North and longitude 37°56' 05' and
38°39' 11' East. The prevailing climate is hot and dry during summer and very cold during winter; with
average rainfall reaches 100mm (20 years average) and high rate of evaporation (Meteorological Dept.
of Jordan, 2002).
The area is considered poor in plant cover since there are no natural forests there. Many farms exist in
this area depend on irrigation water, most of them are used for vegetables production (ground crops)
and the others are cultivated by olive trees. While the herbaceous grazing lands such as Artimisia Alba,
Rattamus and others are rain- fed types. (The Hashemite Kingdom of Jordan, 1991).
Soil in the studied area is derived from limestone rocks or limestone accompanying with basalt rocks in
some areas. The most important characteristics of this soil are high rates of silt and calcium carbonates.
Salinization and gypsum increased towards eastwards. The low level of organic matter and the formation
of the surface cause high rates of erosion by wind and water. The main types of soil prevailing in the
areas below an inclination level of 10% are aridsols and entisols in steeper areas (The Hashemite
Kingdom of Jordan, 1991).
Figure 1: Location of the Studied Area
Figure 2: Methodology of the Study
