Banca de DEFESA: MATHEUS DA COSTA GONDIM

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : MATHEUS DA COSTA GONDIM
DATE: 30/07/2021
TIME: 14:00
LOCAL: Sala virtual no google meet
TITLE:

OPTIMAL NUMBER AND SIZE OF CONGLOMERATES TO ESTIMATE WOOD VOLUME IN AMAZON FOREST

KEY WORDS:

Uncertainty analysis; Monte Carlo method; Sampling theory; Accuracy and precision.

PAGES: 70
BIG AREA: Ciências Agrárias
AREA: Recursos Florestais e Engenharia Florestal
SUMMARY:

Smaller and less sample units (SUs) is in general wanted in forest inventories. This because the number of trees measured in the plots is reduced, turning the fieldwork less costly. This research explores resampling techniques for a set of clusters standardized by the Brazilian National Forest Inventory (NFI). The aim was twofold: to identify (1) the smallest cluster and (2) the smallest sample size that provides the same accuracy and precision as the original sample.  Data from 22 0,80 ha clusters were installed in the National Forest of Bom Futuro, Brazil. Three products were considered: (1) volume of the trees with the diameter at breast height (DBH) ≥ 20 cm, (2) volume of the trees with DBH ≥ 50 cm, and (3) volume of the trees with DBH ≥ 50 cm, with stem quality 1 and 2. In Chapter 1, consecutive reductions from 0.80 to 0.08 ha were carried out in two directions (distal and proximal), giving rise to 20 scenarios. In Chapter 2, the sampling intensity with full-size (0.80 ha each) and small-size (0.56 ha each) clusters was reduced from 22 to 4 clusters. The Monte Carlo method was used as the resampling technique needed to compute the accuracy and precision of the wood volume for every scenario. In Chapter 1, for the three products, findings revealed that 0.56-ha clusters reduced in the proximal direction can accurately and accurately estimate the variable of interest. In Chapter 2, also for the three products, all sample size reductions yielded less accuracy and precision than the original sample (22 SUs). This survey presents two main scientific contributions. First, for a given sampling intensity, it is better to reduce size of SUs than the amount of SUs. This relation has been verified in many variable and sampling spaces, being reported in Cochran (1977). This research confirms, therefore, that same relation for the variable ‘wood volume’ in an Amazonian Forest remnant. The second scientific contribution is to prove statistically that clusters installed in the Amazon could be reduced in size, for volumetry purposes. This size reduction could, inclusive, become the NFI less expensive.

BANKING MEMBERS:
Externo ao Programa - 2295917 - ANGELO AUGUSTO EBLING
Externo à Instituição - EMANUEL JOSÉ GOMES DE ARAÚJO - UFRRJ
Presidente - 2315025 - HASSAN CAMIL DAVID
Externo à Instituição - VINÍCIUS AUGUSTO MORAIS