the 
the 
tal 
and 
was 
ing 
iged 
prid 
mon 
are 
ape 
xes 
are 
can 
ata 
on. 
be 
the 
ing 
are 
ata 
to 
an 
the 
ach 
the 
ing 
tic 
ral 
ons 
ACY 
neighbor locations. 
The location accuracy in remote 
sensing can be expressed as RMSE (Root 
Mean Square Error), which is derived 
from the georeferencing algorithms that 
rectify images to the map coordinates. 
The RMSE is the square root of the 
average of the square of the errors and 
reflects the proportion or maximum or 
minimum pixels number that the control 
points of the image differ from the map 
or from the reference control points. 
Although,the RMSE not always reflects 
the accurate location of all the pixels 
in an image. The RMSE addresses only 
the control points and only related to 
the map. The most accurate way of 
examining the location accuracy i= a 
topographic control with GPS data, which 
has a hig implementation cost. 
The most common way to represent the 
thematic accuracy or classification 
generated by remote sensing is by an 
error. matrix. An error... matrix: is... a 
square arrengement of numbers belonging 
to a particular category related to the 
current category, as the one verifyed on 
the soil. The coiumns usually represent 
the reference data, while the lines 
indicate the classification generated by 
the remote sensing data. An error 
matrix is an effective way to represent 
the accuracy because the accuracy of 
each category is fully described with 
the inclusion and exclusion errors, 
present. in the classification. The 
error matrix can be used as a point of 
departure for a series of statistic 
descriptive and analytics measures. 
The two most common thematic accuracy 
measures use binomials probabilities or 
agreement Kappa ratio. Binomial 
probabilities are based on the correct 
percentage, so that they do not separate 
inclusion and exclusion errors. On the 
contrary, the Kappa ratio gives a 
different measure between the observed 
agreement of two maps and the agreement 
made by chance. 
The Kappa advantages are that its 
calculation considers the elements out 
of the error matrix diagonal and that 
the Kappa conditional ratio can be 
calculated for individual categories. 
In order to standardize procedures for 
reports and for thematic statistical 
maps, the error matrix shall present and 
include the inclusion percentage error 
by category, the exclusion percentage 
error by category, the total correct 
percentage, number of sampled points, 
map accuracy and the statistic Kappa. 
Final Product 
The objective of most investigations 
of remote sensing and Geographic 
Information Systems is to produce a 
product which gives important accurate 
and quick information for scientists and 
administrators. The product may have 
several configurations, including 
thematic and statistic summary. 
The thematic maps may contain 
statistic and dynamic information. A 
statistic  thematic map is produced by 
analyzing the information collected in a 
353 
unique date, while a dynamic map must 
produce the changes occured between 
Succeeding observation dates. 
In. order to reduce the error of the 
final product there are important 
procedures for these maps generation. 
A substantial amount of error can be 
removed if the reader is provided with a 
complete  cartobibliographic citation, 
1.e., the genealogy or lineage of the 
map products. In some remote sensing 
Software packages there are methods for 
tracking the processing flow for a 
particular datafile. The general 
proposition has been to create a 
historical file by Ivsting all the 
operations and parameters that have been 
applicable to a data join. Other kinds 
of error can be reduced by using good 
cartographic design principles in the 
generation of the map products, like the 
legends. 
Geometrical error in final thematic 
map products can be inserted by the use 
of base maps with different grades, 
different national horizontal datum in 
the Source materials and different 
minimum mapping units that are, then, 
resampled for a final minimum mapping 
unit. 
It is commanding the improving of map 
legends that include  cartobibliographic 
information of the geometrical nature of 
the original source material. This is 
the only way to allow the readers to 
judge the geometric accuracy of the 
thematic maps final products. 
The final map must be uniform in its 
accuracy even being the addition of 
information from several sources. It is 
important for this map reader to know 
which of these sources are reliable 
thematic sources. There is a large 
necessity of standardization of the 
project and of the functions of the 
reliable thematic diagrams. 
The fundamental principles of the 
cartographic projects must be followed 
specially in the building of the classes 
interval legends for thematic maps. 
More and more, the remote sensing and 
GIS information are presented in 
electronic viewing device and excessive 
classes intervals and poor colour 
selection variations produce poor 
cartographic communication on the CRT 
visualizers. 
While a lot of progress has been 
done on the statistic thematic maps, 
dynamic changing detection maps almost 
always have poor legends. Too much 
research is necessary to make possible 
the report of the ocurred changes, 
accurately, to the reader. 
Several scientists, nowadays, have 
superposed vectorial images with matrix 
images. This powerful technique gives a 
generic basic map that the reader can 
use to guide and evaluate the  vectorial 
data. Unfortunately, there is no 
standardization in relation to the 
optimum viewing conditions for the 
bottom image or to the optimum project 
of the vectorial data. Researches are 
necessary to standardize and provide, as 
products, thematic maps which 
incorporate a matrix/vectorial