Data - Historical Land Use/Land Cover Classification Using Remote Sensing - page 57

less absolute than ''truth''''. Generally, no study should be carried out without
study area visits, but it is often possible to select training and testing samples for
accuracy assessment from high resolution aerial photographs or from a suitable
thematic maps showing LULC-categories, in addition to topographic maps and
Google Earth data. On the other hand, social science can help to confirm and
analyze remotely sensed observations (e.g., validating remote observations versus
data gathered on the ground). As, it is possible to determine a number of land use
activities (classes) during classification remote sensing imagery based on a few
social behaviors (McCoy 2005 ). One of the interesting example presented by
Lesschen et al. ( 2005 ) presents other data sources such as questionnaires. It is
particularly helpful to obtain management-related data (e.g., agricultural crop
cycles) and can also give insight into the main factors of LULC-change.
4.2.1 Field Reference Data
One of the mainly steps in any remotely sensed based study is gathering the
thematic data (attribute data, such as qualitative breaks by vegetation cover den-
sity: low, medium, high), and measurements (e.g., a quantitative differentiation of
vegetation cover density by break points: \10, 11-40, [40 %), or observations
(e.g., determination which category of the legend is more suitable to a surface
feature) of the phenomenon of interest in the field. This is also the most difficult
step, because it is a very time-consuming, often boring task which entails diffi-
culties such as what want the researcher measuring and observing, where it is
important to determine that before the field work, and then it is important to
choose the method to be applied in the field to gather the required data. Ground
truthing is important for remote sensing to properly identify objects, provide
precise image registration and verify results. Before beginning to gather reference
data in the field, two steps must have been completed: (1) study goals must be
obviously determined; and (2) a classification scheme for all LULC-classes must
have been selected (McCoy 2005 ).
Spectro-radiometry is a frequently used ground-based reference data source in
remote sensing techniques. It can measuring the values of radiance, irradiance,
reflectance or transmission of individual targets or objects, by locating the radi-
ometer above the targets of interest, and records these values as digital spectral
quantitative records. It used mostly as hand-held cameras (or mounted on a tripod,
tower, tractor etc.). After finishing the measuring process, the user compares them
to the biological, chemical and physical characteristics of the object. For agri-
culture, red and NIR portions of the EMS profile utilized particularly to calculate
and generate the Vegetation Indices (VI) that are correlated with parameters of
canopy structure (e.g., LAI). Spectro-radiometers named also radiometers or
InfraRed Thermometers (IRT) (Schowengerdt 2007 ).