Relação entre transparência da água obtida in situ e a reflectância da imagem landsat 5 tm do reservatório Passo Real

Abstract

The water quality monitoring is generally done from in situ sampling. However, the financial operational cost is high, beyond the difficult sampling in different sectors of the hydric body. As a possible solution for this problem, it can be used the historical basis of Landsat 5 TM images that allow the realization of space-temporal studies of aquatic environments. In this context, the present research proposes to verify the relation between water transparency data identified in situ and the values of spectral reflection of the Passo Real Reservoir obtained through Landsat 5 images. The water transparency was measured in situ with the use of a Secchi disc in 31 sample points of the reservoir. The satellite images used for the temporal analysis of the water reflection were produced by the Landsat 5 TM satellite, from which it is possible to extract the pure pixels related to the spacial resolution of 30 meters. The image selection was from the scene availability without the presence of clouds in the dates closer to the field work realization. The image processing for the correction of atmospheric effects and the transformation of digital numbers into reflection values were done with the ENVI 5.0 software. The spacings of the transparency data and reflection values were done by the interpolation in the Spring 4.3.3 software. In the temporal analysis from March 2009 to July 2009 the measures of transparencies of the Passo Real Reservoir in its major were superior to 100 centimeters. The periods from August 2009 and October 2009 presented transparencies inferior to 100 centimeters and of shorter data variance in relation to the ones studied before. The reflection data in the space-temporal context indicates that the main alterations correspond to the tributary sectors of the Passo Real Reservoir. The spectral data of the Band 3 of the TM sensor were related with the water transparency variable in the 31 sample points. The models generated by linear regression showed that the water transparency have relation with the reflection data in the red region. the models with the best adjustment were obtained for the data from March 2009 to July 2009, period with indexes of higher extent and variability of transparency and spectral reflection data, although all periods have presented meaningful regression in a trust interval of 99,9% (n=31). The presence of photosynthetic pigments in the water diminish the level of correlation of transparency and reflection data in the red strip. This fact occurs due to the Band 3 of the TM sensor be sensible to electromagnetic energy absorption by photosynthetic pigments. The satellite images characterize themselves in important tools for the comprehension of the space-temporal functioning of the reservoir.