Research Methodology - Historical Land Use/Land Cover Classification Using Remote Sensing - page 86

Fig. 5.15 Excursion-GPS-points-2007 and other GPS-measurements from ICARDA and GORS

grow without problems. A second campaign was conducted in July 2009 for
complementary information and some GPS-measurements based on the knowledge
of the farmers. These two field work periods were held to increase the under-
standing of the patterns of LULC in the study area. Preliminary image classifi-
cation (unsupervised) and RGB-composite imagery of the study area were printed
to show target areas to be surveyed depending on the accessibility of each site. The
data were gathered from different sites depending on the differing soil types and
irrigation systems in the study area. Random sampling methods were used. Each
plot was registered by using GPS-technology (using a GARMIN-Colorado-300
global positioning receiver) to allow for further integration with the spatial data in
a geographic information system (GIS) and image classification programs.
Information was gathered based on specific procedures such as: identification of
the dominating species of trees, shrubs and herbs; detection of the physical aspects
of the soil; conduction of interviews and group discussions with local farmers to
extract historical information about the LULC in the study area; and gathering of
information about prior LULC activities regarding to types, densities, distribu-
tions, and species.
Figure 5.16 illustrates one example from the study area as explanation of the
steps followed during the excursion in 2007 to collect ground truth data (especially
for agriculture). The outputs of this first experimental stage were the gathering of
the training samples and the testing of sites for automated supervised classification