Banca de DEFESA: MARCELA LOPES LÁZARO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
DISCENTE : MARCELA LOPES LÁZARO
DATA : 30/08/2018
HORA: 13:30
LOCAL: Sala 24 Departamento de Solos do Instituto de Agronomia da UFRRJ
TÍTULO:

Prediction of soil carbon, nitrogen and phosphorus by Vis-NIR reflectance spectroscopy

PALAVRAS-CHAVES:

near infrared, organic agriculture, organic matter.

PÁGINAS: 60
GRANDE ÁREA: Ciências Agrárias
ÁREA: Agronomia
SUBÁREA: Ciência do Solo
ESPECIALIDADE: Manejo e Conservação do Solo
RESUMO:

Vis-infrared (Vis-NIR) reflectance spectroscopy has been used as a method to predict soil attributes, whether chemical, physical, biological or mineralogical, with fast, non-destructive results when compared to conventional methods routine analysis. This work was carried out in a Planosol Haplic in the Integrated System of Agroecological Production (SIPA) in Seropédica - RJ, in an experimental module of intensive organic production of vegetables for prediction of carbon (C), nitrogen (N) and phosphorus reflectance spectroscopy. A total of 294 soil samples were collected at 20 cm depth for chemical characterization of C, N and P and spectral readout on the Field Spec 4 spectroradiometer. The spectra for each of the soil samples were generated and quantitatively analyzed by regression analysis. (PLS) and validated for two sets of data. In parallel, five preprocesses were tested to prepare the spectra for the purpose of improving predictions. For the validation of the models, the values obtained for C prediction were: R² between 0.69 and 0.75; RMSE and RPIQ ranging from 0.12 to 0.14 and 2.21 to 2.59, respectively. For N, the R² values obtained ranged from 0.62 to 0.77; the RMSE maintained the same approximate values at 0.015 and the RPIQ ranged from 2.32 and 3.03. For P, values obtained ranged from 0.47 to 0.57; 33.40 to 38.47 and 2.06 to 2.54 for R², RMSE and RPIQ, respectively. The results suggest that Vis-NIR reflectance spectroscopy is a promising technique for predicting soil carbon and nitrogen. The models adjusted with the gross reflectance data and pre-processed by smoothing presented better predictions of these attributes. On the other hand, soil phosphorus did not present prediction models with good fit, for none of the pre-processing tested.

MEMBROS DA BANCA:
Externo à Instituição - FABRÍCIO SILVA TERRA - UFVJM
Externo à Instituição - GUSTAVO DE MATTOS VASQUES - EMBRAPA
Interno - 1220296 - MARCOS BACIS CEDDIA