il 2004
vation
ym the
ing as
RUTA
astline
3S over
RDAS
or the
ninary
? Steps
ialysis
spatial
or the
cening
esides
of the
nap in
id.
of the
ion of
rt and
eters,
ystem
e data
ysical
of fast
spatial
d map
2
TA)
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
: 2000 (Lando ETM
rn E 199? (Ludei TM;
|| 77 1990 (Lande
^ 1580 (Mag)
Scara 1 : 5 606
Figure 4. Shoreline evolution over 20 years as extracted from
Landsat data.
3.1. The changes of the Romanian costal zone
The Rumanian coastal zone can be divided from the
geomorphological point of view, in two main units. The
northem unit and the southern unit. In the northern zone prevails
the low relief, characteristic to the Danube Delta, with narrow
blokes. Transported by the strong currents of that zone, the
sands built a barrier in the Sf. Gheorghe zone. These barriers
closed the Razelm Lake. The sediments form the modern times
are made mainly from quartz sand (7096 silicon). Heavy metals
do not go over 3%. The transport of sediments that come from
the regions north to the Danube Delta have higher silicon
content, of almost 9096. The evolution of the delta zone started
during the quarternary and was strongly influenced by the
modifications into the sea level of that time.
The Delta was formed during the period of the sea level retreats
through the alternate developing of the river Danube, each
developing it's own deltaic structure. Presently there are three
Danube branches active and only Chilia is still developing it's
own deltaic structure. The other coastal sectors are retracting,
being influenced by the decrease in the Danube sediments over
the last century. The long term studies shown a decrease of the
Black Sea level of almost 2.5mm/year in the Vama Veche
region, while other measurements evidentiated the rising of the
sea level of 1.2 to 1.8 mm/year at Sulina and almost 3.3mm/year
at Constanta. The ground in the delta zone is going down with
1.3-2mm/year because of the sediments phenomenon and the
zone's tectonics. The tide phenomenon in the Romanian coast
not is easily detected because of other fluctuations. The other
fluctuations of the Black Sea level are caused by the dynamics
in the river's debits that go into the Sea, modifications of the
water flow in the Bosfor straits and the precipitations/
evaporations relation in that zone. One of the most important
factors that influence the hydrological budget of the Black Sea
is the volume of water, which goes through the rivers that form
the hydrological pool of the Black Sea. The Danube River has
the highest volume between April and June. The building of
many dams over the Danube River and it's affluents
considerably reduced the transport of sediments.
Another important factor is the winds and waves regime. The
average wind speed in the NV region of the Black Sea is
between 6.5 and 5m/s. The main directions of the wind is N,V
and S, a greater weight having the NV direction. During the
summer months the predominant direction is S-SE. The storms
have a predominant N direction, with an average wind speed of
9.8m/s, during a period of time of 8 to 22 hours. There is a 5096
probability that aver one year to have waves higher than 0.2m.
The retreat of the coastline with almost 10m/year is more
obvious along the Sakhalin barrier and south to Ciotca, and also
between Portita and Chitiuc. Then the coastline advances
immediately south of Sulina where it is accompanied by the
presence of some very shallow waters. See figure 5
781
4. Sources of error in change detection
A first approach to classify errors in environmental change
analysis with remote sensed data is dividing the sources of error
in instrumental errors and method errors. A more detailed
description of errors can include the data acquisition errors, data
processing, data analysis and data conversion errors .
In the process of error assessment, several errors can occur:
positional errors, registration, differences, data entry error for
reference data, interpretation and delineation of reference data,
reference data and remote sensed data are not simultaneously
collected, classification errors. The results are affected by
different errors, at different levels and the positional and the
thematic information obtained from the two data sets is not of
the same precision. From the merging of the two data sets will
result an unknown precision of the final product.
4.1 Positioning error
The resulted thematic classification assigned to each identified
class a spatial location on the image. During this process a
generalization operation is performed. The class boundaries are
also affected by misclassification of the marginal pixels.
Horizontal accuracy for map products at scales greater than 1:20
000 must be less than 1096 of tested points to have a greater
error of 0.85cm, measured at the map's scale. For maps at scales
1:20 000 or smaller, the admissible error is 0.51 mm. The
horizontal components are defining standard position error in
which are contained 9096 of point coordinate discrepancies.
Another accuracy criteria is the map standard deviation.
1
A
Ya, s ~~. 1857
| 1910
1962
v 2008 =
Bratul Chilia al e
&
©
S ; Ty
Wim NN
A i910
«Hu
: 2002
u
LEGEND
+ Lp noie ia vet
non 1887
---1910
— 1962
— 20603
* R62 Control points 1962
SCALE
6 $ 10 15 20km
mmi —dmeni— 3i
Figure 5. The dynamics of the littoral corridor between year
1857 and 2003.
Standard deviation for tested points must not have a value over
the value calculated :
0,5
d= 5 | (1)
i=11"
where d = standard deviation
li= point error
n = number of points
