@InProceedings{MoreiraTeixGalv:2015:AnQuCo,
author = "Moreira, Luis Clenio J{\'a}rio and Teixeira, Adunias dos Santos
and Galv{\~a}o, L{\^e}nio Soares",
affiliation = "{} and {} and {Instituto Nacional de Pesquisas Espaciais (INPE)}",
title = "An{\'a}lise quantitativa da concentra{\c{c}}{\~a}o de sais nos
solos a partir de dados de espectroscopia de reflect{\^a}ncia",
booktitle = "Anais...",
year = "2015",
editor = "Gherardi, Douglas Francisco Marcolino and Arag{\~a}o, Luiz
Eduardo Oliveira e Cruz de",
pages = "3919--3926",
organization = "Simp{\'o}sio Brasileiro de Sensoriamento Remoto, 17. (SBSR)",
publisher = "Instituto Nacional de Pesquisas Espaciais (INPE)",
address = "S{\~a}o Jos{\'e} dos Campos",
abstract = "The aim of this study was to evaluate the possibility of using
reflectance spectroscopy to quantify the salt concentration of the
soil enabling the use of hyperspectral imagery for mapping large
degraded areas. To develop statistical models 93 soil samples, and
to calibrate 2/3 and 1/3 were used to validate. In the two sample
subsets spectral measurements were made in the laboratory using
the spectroradiometer FieldSpec Pro under controlled conditions
and measurements of electrical conductivity (EC) were performed.
Three statistical models were used to analyze the reflectance vs
EC: linear regression, normalized salinity index (NDSI) and
partial least squares regression (PLSR). The linear model was
developed to better results with the band positioned at 1945 nm
showing significant predictive power (R2 = 0.50, RMSE = 0.987 and
RPD = 1.47). Still, it was lower than the model developed from the
NDSI (using 1875 and 1935 nm with R2 = 0.836; RMSE = 0.54; RPD =
2.44). Two PLSR models were constructed: one using all the
spectral information (PLSR1) and other bands without atmospheric
interference (PLSR2). The PLSR1 showed better results (R2 = 0.883,
RMSE = 0.44 and RPD = 2.90) compared to the other models developed
in this work suggest that the greater the number of spectral
information used in the modeling, the greater the ability to
predict. However, the optimization of the number of variables to
compose the predictive models may be made, enabling better results
with the least number of possible input variables.",
conference-location = "Jo{\~a}o Pessoa",
conference-year = "25-29 abr. 2015",
isbn = "978-85-17-0076-8",
label = "780",
language = "pt",
organisation = "Instituto Nacional de Pesquisas Espaciais (INPE)",
ibi = "8JMKD3MGP6W34M/3JM4C8L",
url = "http://urlib.net/ibi/8JMKD3MGP6W34M/3JM4C8L",
targetfile = "p0780.pdf",
type = "Sensoriamento remoto hiperespectral",
urlaccessdate = "27 abr. 2024"
}