Comparação De Métodos E Processos De Amostragem Para Inventário Em Floresta Ombrófila Mista

Abstract

RESUMO ABSTRACTInventários florestais são indispensáveis para avaliar quantitativa e qualitativamente os recursos existentes em determinada propriedade ou região. Diferentes métodos e processos aplicados aos inventários repercutem em estimativas com maior ou menor precisão. O objetivo deste estudo foi comparar as dimensões de unidades amostrais, sua distribuição na população e intensidades amostrais na estimativa da densidade, área basal e volume por hectare em um fragmento de Floresta Ombrófila Mista no Sul do estado do Paraná. Foram analisadas 28 configurações de amostragem, as quais foram avaliadas por meio dos respectivos erros de amostragem relativos em uma área submetida ao censo e comparadas a partir do erro real relativo. Foram testadas duas frações de amostragem, correspondendo a 8 e 16% da área da população, considerando os processos aleatório e sistemático. Foram avaliados oito tamanhos de unidade amostrais de 400 a 1.050 m 2 . As configurações amostrais avaliadas resultaram em erros relativos inferiores a 20% para as três variáveis, valores esses considerados aceitáveis para florestas nativas. Os menores erros reais relativos foram encontrados para unidades amostrais de 1.000 m 2 para as três variáveis estudadas, variando apenas na sua forma, sendo 25 m x 40 m para o número de árvores por hectare e 10 m x 100 m para área basal e volume por hectare. Concluiu-se que maior fração amostral gera estimativas mais precisas para número de ár-vores por hectare e área basal. A forma das unidades amostrais influenciou os resultados.Forest inventories are essential to evaluate quantitatively and qualitatively the existing resources in a given property or region. Different methods and processes applied to inventories provide estimates with more or less precision level. The aim of this study was to compare different sizes of sample units, their sampling design, and sampling intensity to estimate tree density, basal area and volume per hectare in a remnant of Araucaria Forest in southern Parana State. We tested 28 sample design, which were evaluated by the respective sampling errors in an area under census. and compared by the relative real error. We compared two sampling areas, corresponding to 6 and 16% of sampling fraction, considering the random and systematic sampling designs. Eight sample unit sizes were evaluated, ranging from 400 to 1.050 m 2 . All the tested sample treatments resulted in a relative error less than 20% for the three variables analyzed, which is satisfactory for native forests. The lowest real relative errors were found for sample units of 1.000 m 2 for the three variables, varying only in their form, being 25 m x 40 m for tree density and 10 m x 100 m for basal area and volume per hectare. The larger the sample area the more accurate were the estimates. It was concluded that larger sampling fraction results more accurate estimates for tree density and basal area. The sampling design influence the estimates.PALAVRAS-CHAVE: Amostragem, Araucária, Censo florestal.

EFFECTIVENESS OF SAMPLING METHODS FOR THINNED AND NON-THINNED Pinus taeda L. PLANTATIONS IN THE MOUNTAIN REGION OF SANTA CATARINA, BRAZIL

“…It was found that the relative sampling error in all estimates performed was higher than the real error of the estimate. This feature was also observed by Sydow et al (2017) when comparing the sampling error with the census of a mixed ombrophilous forest fragment; these authors observed an overestimation of the sampling error, as in this work. It is also noteworthy that the sampling error reveals the probability that the sampling can adequately represent the population, but does not ensure the accuracy of the forest inventory estimates.…”

Artificial intelligence tools and vegetation indices combined to estimate aboveground biomass in tropical forests

“…66 Low NDVI values are related to disturbed forests by logging activities and wildfires that affect canopy cover. 17,64,67 High NDVI values are commonly observed in undisturbed forest sites, [67][68][69] as observed in this study within DLOF. The high correlation result between vegetation indices and AGB observed in our analysis is accordance with the available scientific literature.…”

“…42 Based on the high complexity of the studied variables and the high variability commonly observed in tropical forests, we considered an error limit of ∼20% acceptable, which indicates an achievement of robust prediction results. 26,68 The ANN has been applied by several studies in forest environment showing high predictive power compared to classical regression models for estimating variables of interest. 21,26,27,75,76 In addition, the ANNs and remotely sensed data combined can provide higher modeling precision in forest sites than the regression models, as they are able to assimilate a high complexity and variety of vegetation, environment, and climatic aspects.…”

“…Based on the high complexity of the studied variables and the high variability commonly observed in tropical forests, we considered an error limit of ∼20% acceptable, which indicates an achievement of robust prediction results 26 , 68 . The ANN has been applied by several studies in forest environment showing high predictive power compared to classical regression models for estimating variables of interest 21 , 26 , 27 , 75 , 76 .…”

Precision and accuracy of sampling methods in thinned Pinus taeda L. stands

“…Among the known sampling methods, the fixed-area method is the oldest and most widespread one in Brazil; the selection of trees is made according to the frequency in which the trees are arranged in the forest and to the size of the sampling unit (SANQUETTA et al, 2014). This method is widely cited in the forest literature and considered, in several cases, as a comparator in different studies, as in the works of Moscovich et al (1992), Druszcz et al (2010), Nakajima et al (2011), Santos et al (2013, Téo et al (2014), Miranda et al (2015), Sydow et al (2017), andOliveira et al (2019).…”