Banca de QUALIFICAÇÃO: BLENDA PEREIRA BASTOS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : BLENDA PEREIRA BASTOS
DATE: 28/11/2022
TIME: 10:00
LOCAL: Departamento de Petrologia e Geotectônica
TITLE:

AIRBORNE GEOPHYSICAL DATA IN PREDICTIVE 1 MODELING OF
2 SOIL ATTRIBUTES IN BOM JARDIM -RJ

KEY WORDS:

Geophysics, predictive modelling, Bom Jardim de Minas

PAGES: 27
BIG AREA: Ciências Exatas e da Terra
AREA: Geociências
SUBÁREA: Geologia
SPECIALTY: Geologia Ambiental
SUMMARY:

Geophysical data have great potential to represent soil-forming factors, such

6 as parent material and relief, aiding the prediction of soil attributes in digital soil mapping

7 through machine learning algorithms. The research paper goal was to apply predictive modeling

8 techniques using airborne geophysical data, terrain and hydrologic covariates obtained from the

9 Digital Elevation Model (DEM) aiming to assess the importance of those covariates to modeling

10 soil properties and evaluate tropical landscape dynamics. The study was carried out in Bom

11 Jardim County, Rio de Janeiro, Brazil, with a database consisting of 208 superficial soil samples

12 and a total of 37 covariates. Non-explanatory covariates for the selected soil attributes were

13 previously excluded using nearZeroVar, findCorrelation e rfeControl methods. Through the

14 selected covariables, the Random Forest (RF) and Gradient Boosting Machine (GBM) models

15 were performed with separate samples for training (70%) and validation (30%). The model’s

16 performance was evaluated quantitatively through the coefficient of determination (R2) and root

17 mean square (RMSE). From RFE analysis geophysical data composes more of 50% of the most

18 important covariates for sand. For clay, the contribution of these data was smaller, comprising

19 35% (rfFuncs) and 41.66% (caretFuncs). The best results were obtained by the RF model with

20 R2 and RMSE values equal to 0.14 and 156.24 (g/kg), respectively, for Sand (g/kg) and R2 and

21 RMSE values equal to 0.10 and 132.42 (g/kg), respectively, for Clay (g/kg).

COMMITTEE MEMBERS:
Interna - 1767261 - OLGA VENIMAR DE OLIVEIRA GOMES