Banca de DEFESA: NIRIELE BRUNO RODRIGUES

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : NIRIELE BRUNO RODRIGUES
DATE: 26/11/2020
TIME: 13:00
LOCAL: vídeo conferência
TITLE:

Establishment of Heavy Metal Quality Reference Values and
Digital Mapping of Soil Chemical and Physical Attributes of the North and Northwest
Fluminense, RJ.

KEY WORDS:

Natural heavy metal contents; Soil pollution; Digital soil mapping

PAGES: 110
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUBÁREA: Ciência do Solo
SUMMARY:

The demand for information about the soil resource, especially the contamination by heavy
metals of agroecosystems has increased in recent decades as a result of the massive use of
agrochemicals and animal production waste, which has caused the accumulation and transfer
of toxic metals to food, restricting soil ecosystem services and offering risks to human health.
The North and Northwest regions have great socioeconomic relevance in the agricultural sector
and in the oil activity. It presents heterogeneous physiographic characteristics, demanding
support information for soil management and conservation for sustainable agricultural
production. Thus, in search of reference values for heavy metals, associated with the elucidation
of mechanisms and the spatial distribution of chemical and physical attributes in soils of
dynamic and heterogeneous environments, this research aimed to establish reference values for
soil quality for heavy metals: As; Pb; Cd; Ni; Cu; Co; Ba; Cr; Zn; Mn, and Al, and to spatialize
the contents of these metals associated to environmental covariables, from the North and
Northwest regions of the State of Rio de Janeiro. For the methodological procedures of Chapter
I, the use of descriptive statistics associated with the technique of digital soil mapping was
adopted, with the help of the software RStudio (3.6.1), Saga GIS (2.1.2) and Quantum Gis (v.
3.4). For terrain morphometric variables, MDE-HC and Landsat 8 sensor data were used,
totaling 21 predictor environmental covariables with spatial resolution of 90 m. For the
sampling procedure, the Latin Conditioned Hypercube (cLHS) method was used, collecting
samples in areas with low or no anthropic activity. We used data only from the superficial layer
(0- 20 cm), in 97 sample points. For the prediction of chemical and physical attributes the
Random Forest (RF) model was adopted, implemented via RStudio. The results indicated that
for the mean quadratic error parameter (RMSE) a variation between 1.32-8571.34 was obtained,
so the data set also showed significant changes. In chapter II, the results obtained via the
VarImport ranking found that, in comparison with the Landsat-8 image index, the covariates
from MDE-HC obtained a better performance to predict soil attributes. The coefficient of
determination (R²) of the models of heavy metals and soil texture varied between 0.22-0.87, Pb
(0.55), Ni (0.62), Co (0.66), Mn (0.73), Cu (0.74), Al (0.77) and Zn ( 0.77) and As (0.86)
considered strong. Regarding the granulometric fraction, the variability of clay (0.87), sand
(0.84) and silt (0.85) is classified as strong, showing different patterns of variability and use of
similar predictive covariates. For chapter 2, the same sample mesh of 97 points was also used,
considering 2 depths (0-20 cm and 20-40 cm), totaling 194 soil samples. For the determination
of the heavy metal pseudo total contents, the 3051A (USEPA) method was used, and the extract
readings were performed by ICP-OES. For the statistical treatment, the technique of
multivariate statistics was used to establish the reference values of soil quality. The samples
were grouped in 3 groups those of group 1 (G1) with the highest reference values and those of
group 3 (G3) with the lowest. The results obtained were satisfactory through the integration of
geochemistry with spatial analysis, contributing to fill the scientific gap about the VRQS for
the North and Northwest regions of Rio de Janeiro, thus corroborating as support for
environmental legislation in the state of Rio de Janeiro, and also for the knowledge and
understanding of the natural levels of heavy metals associated with spatial variability and their
interaction with soil properties.

BANKING MEMBERS:
Externo à Instituição - FERNANDA ARAÚJO DOS SANTOS - UNICAL
Interna - 2223668 - HELENA SARAIVA KOENOW PINHEIRO
Presidente - 387263 - NELSON MOURA BRASIL DO AMARAL SOBRINHO