Banca de QUALIFICAÇÃO: DOUGLATH ALVES CORRÊA FERNANDES

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : DOUGLATH ALVES CORRÊA FERNANDES
DATE: 14/04/2023
TIME: 09:00
LOCAL: VIDEOCONFERÊNCIA
TITLE:

Digital mapping of the chemical attributes of the soil in the oil and gas production area in the northern region of the Recôncavo Baiano. Case study of the semi-arid region - Sátiro Dias, Bahia.

KEY WORDS:

machine learning, digital soil mapping, Randon Forest

PAGES: 30
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUMMARY:

In Brazil, the execution of soil mapping with appropriate scale throughout the national territory is a permanent demand of research institutions and planning agencies. The lack of this information is common, as there are several limitations to the acquisition of soil data or its attributes, including the high cost of soil surveys, storage of samples, and laboratory processing. In this sense, detailed knowledge of soil properties at different scales is an urgent need, in order to help land managers make spatially explicit decisions about soil management and conservation, rational land use planning, prediction of future scenarios, and serve as a source of data for spatiotemporal modeling. In this context, digital soil mapping emerges as an alternative to increase the viability of soil surveys, using interactions between soil properties and environmental covariates that would ultimately allow optimization of the mapping of such properties in the Brazilian semi-arid region. Therefore, the objective of this work is to model and map the soil chemical attributes at different depths using machine learning algorithms for a study area located in the municipality of Sátiro Dias, in the northern region of the state of Bahia, Brazil. This study considered a smaller area of approximately 200 km², which is representative of a larger area (900 km²) and is therefore a reference area (RA). Thus, the hypothesis of the work considers that there is representativeness of soil types and attributes in both regions mentioned above, and therefore it is possible to model the target attributes in an RA and then extrapolate this model to a larger area, i.e., an external area (EA).

COMMITTEE MEMBERS:
Externo à Instituição - ALEXANDRE TEN CATEN - UFSC
Externo à Instituição - GUSTAVO DE MATTOS VASQUES - EMBRAPA
Presidente - 1220296 - MARCOS BACIS CEDDIA
Externo à Instituição - WALDIR DE CARVALHO JUNIOR - EMBRAPA