Banca de QUALIFICAÇÃO: ANA LIVIA ARAUJO DE AZEVEDO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : ANA LIVIA ARAUJO DE AZEVEDO
DATE: 10/04/2025
TIME: 14:00
LOCAL: Google Meet
TITLE:

Raimundo Nonato Agricultural Reserve soil survey: integration between conventional and ethnopedological approaches

 

KEY WORDS:

soil mapping; pedometrics; etnopedology; rural settlements.

PAGES: 112
BIG AREA: Ciências Humanas
AREA: Geografia
SUMMARY:

The inclusion of large-scale water projects in the Northeastern backlands brings a complex set of
benefits and challenges: new regional socioeconomic dynamics and relocation of communities
affected by dam projects. In the Raimundo Nonato Rural Settlement, a settlement for rural worker

families affected by the construction of the Oiticica Dam in the municipality of Jucurutu (RN), small-
scale farming and livestock farming are the main economic activities developed. In this context,

knowledge about the support resource for agricultural practices is essential for small farmers who
are establishing and developing new experiences in the settlement area. This study, aiming to
optimize knowledge about the soils of the Raimundo Nonato Settlement, considers two approaches:
Ethnopedology, integrating the traditional knowledge of the community and academic knowledge;
and soil survey and mapping that provides a diagnosis of the soils and their main characteristics,
facilitating the effectiveness of sustainable agricultural practices. The objective of this work is to
analyze the spatial distribution of soils in the settlement, using the pedological approach in
conjunction with a hybrid soil mapping methodology. To carry out the work, regarding the soil survey
stage, conventional and digital survey approaches were used, at a semi-detailed level (1:25,000).
Observation and sampling points were generated using the Conditioned Latin Hypercube method,
fed with geomorphometric covariates extracted from a Digital Terrain Model, obtained by LiDAR
technique, used together with machine learning algorithms providing predictions about the
spatialization of soil attributes, which will assist in later stages of delimitation of mapping units. The
soil samples were analyzed in the laboratory, for physical and chemical characterization and
subsequent classification in the Brazilian Soil Classification System. In the Ethnopedological
approach, semi-structured interviews will be conducted with the settlers using the snowball
sampling method, seeking to understand their perceptions about the soils and the local landscape.
Preliminary results indicate the occurrence of Luvissolo Háplico, Gleissolo Háplico, Neossolos
Flúvicos, Neossolos Litólicos and Neossolos Regolíticos classes, with a predominance of sandy,
eutrophic soils, but with relatively low nutrient content and moderately acidic pH. Random Forest,
Support Vector Machines, Gradient Boosting Machines and Cubist algorithms were used to predict
the data, modeling attributes of coarse fraction, pH, K and available Na. The coefficient of determination (R2) and the performance metrics MAE and RMSE showed variations between the
models for each of the attributes.

 

 

COMMITTEE MEMBERS:
Externo à Instituição - DAVI FEITAL GJORUP
Presidente - 1818226 - JOAO SANTIAGO REIS
Interna - 1726169 - SARA FERNANDES FLOR DE SOUZA
Externa à Instituição - SIMONE CARDOSO RIBEIRO - URCA