Vol. XXXVIII, Part 7B 
In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
11 
[N THE STATE 
os Adami 1 
tal. 
e actions that harm the 
ental protocols between 
f Sao Paulo requires the 
are used to monitor and 
esting method (with or 
le methodology utilized 
harvest type was also 
the 2006/07 season to 
or harvest in the state is 
?ion of the state had the 
tute for Space Research 
'paciais - INPE), the 
ICA), the Center for 
cs (CEPEA) of the Luiz 
JSP) and the Center for 
naintained the Canasat 
Using remote sensing 
s, the Canasat project 
Initially, mapping was 
o (Rudorff et al., 2005), 
n extended to the other 
of Brazil (Rudorff and 
esponsible for 72.7% of 
tate of Sao Paulo, is to 
>r without burning the 
; 2006/07 crop season, 
is been utilized by both 
¡. Beginning with the 
l the type of harvest are 
4A of Sao Paulo State, 
nine if the straw burning 
t the methodology of the 
oring of the harvesting 
Df the harvest areas with 
)6/07 to the 2008/09 crop 
:ct may serve as a basis 
.ra et al., 2005), carbon 
, public health studies 
olicy in the agricultural 
o allow evaluating the 
> not harvested due to 
issential information for 
Figure 1. Location of the state of Sao Paulo within Brazil as 
well as the area containing harvestable sugarcane in the 2008/09 
crop year. 
The identification of the harvesting method was performed by a 
visual interpretation of TM (Thematic Mapper) sensor images 
taken from the Landsat-5 satellite. In the case of cloud cover on 
the TM images, CCD (Charge-Coupled Device) sensor images 
taken from the CBERS-2 and CBERS-2B satellites were used as 
an alternative (Epiphanio et al., 2007). For each orbit point of 
the TM and CCD sensors, a database was created with the data 
of interest from the images obtained by the two sensors. All 
images were registered based on the orthorectificated mosaics 
from TM/Landsat-7 images obtained by NASA (NASA,2007) 
utilizing a first degree polynomial and nearest neighbor 
interpolation. 
Monitoring of the harvest type is only possible after producing a 
map of available sugarcane for harvest. This is then utilized as a 
mask for the remote sensing images and allows monitoring only 
the sugarcane areas that available for harvested in the current 
crop year. This map is prepared by the Canasat project at the 
beginning of each crop season. 
2. METHODS 
Identification of the harvesting method, either burning or not 
burning the sugarcane straw, is currently performed in the state 
of Sao Paulo, the largest producer of sugarcane in Brazil. Sao 
Paulo is located in Southeastern Brazil and has an area of 
248,209 km 2 . Figure 1 shows the location of Sao Paulo State 
and the area of sugarcane available for harvest in 2008/09. 
1) Sep. 10 th , 2008 
4a) May 24 th , 2009 
2) Sep. 10 th , 2008 
4b) Jun. 09 th , 2009 
In contrast with other agricultural crops, sugarcane has a long 
harvest season, lasting from April to December. The remote 
sensing images allow to identify the harvesting method, either 
burning or not burning the sugarcane straw, because the areas 
where sugarcane is harvested after burning present dark tones in 
response to soil exposure (Stoner and Baumgardner, 1981). 
Areas harvested without burning present bright tones because 
the ground is covered by dry leaves (Figure 2.1) (Aguiar et al., 
2009). In Figures 2.5 and 2.6, field photos of recently harvested 
areas without straw burning and with straw burning, 
respectively, can be seen. Both the accumulation of straw after 
harvest and soil exposure from burnt straw may be observed in 
these figures. 
Over time, the correct identification of the harvest method 
becomes less clear. Both weather and post-harvest agricultural 
practices such as the burning of straw in the field after harvest 
are the major factors that affect this identification (El-Hajj et al., 
2009). Figure 2.2 illustrates an area with plots harvested on 
different dates. The difference in time and the use of different 
post-harvest agricultural practices create changes in the color 
and characterization of these plots, (which are differentiated in 
the image). However, all plots were harvested without burning. 
Figure 2.3a shows an image acquired in July of 2008 in which it 
is possible to observe an area harvested without burning. This 
same area, in September of 2008, possesses dark tones due to 
either the straw being burnt following the harvest or due to the 
soil being exposed (Figure 2.3b). Therefore, the less time that 
has elapsed between the harvest time and the image acquisition, 
the more likely it is to correctly identify the harvesting method. 
Declivity is a limiting factor for mechanical harvest. For this 
reason, in many crop fields, harvesting is performed using both 
methods. In the part of the field where the declivity is over 12%, 
burning is still used; however, in the part of the field with lower 
declivity the mechanical harvest is performed. Figures 2.4a and 
2.4b illustrate this situation in a sequence of two dates; in the 
dark plots, with a high declivity, a manual harvest was 
performed (after burning), and in the light areas, with a declivity 
of less than 12%, the sugarcane was harvested mechanically 
(without burning). It should be noted that Figure 2.4b shows the 
presence of some clouds. 
3b) Sep. 10 th , 2008 
6) Jun. 03 rd , 2009 
3a) Jul. 08 th , 2008 
5) Jun. 02 nd 2009 
Figure 2. Temporal sequence of TM/Lansat-5 images, color composition 4(R)5(G)3(B), illustrating different harvest types ( 2.1 and 
2.2), the change in harvest characteristics caused by post-harvest agricultural practices (2.3a and 2.3b), different harvest types due to 
declivity (2.4a and 2.4b), and field photos of recently harvested areas without straw burning and with straw burning (2.5 and 2.6).