Volumetric prediction through punctual kriging reduces sampling effort in pre-cut forest inventories

• Autor(a) principal: Dantas, Daniel
• Data de Publicação: 2020
• Outros Autores: Pinto, Luiz Otávio Rodrigues, Gonçalves, Anny Francielly Ataide, Terra, Marcela de Castro Nunes Santos, Calegario, Natalino
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Caderno de Ciências Agrárias (Online)
• Texto Completo: https://periodicos.ufmg.br/index.php/ccaufmg/article/view/15927
• topic: Semivariograma; Inventário Florestal; Intensidade amostral; Semivariogram; Forest inventory; Geostatistical model

The objective of this work was to evaluate the performance of a geostatistical estimator to estimate the wood volume per hectare of an eucalyptus stand, considering different sample intensities. The database was derived from pre-cut forest inventories, with a sampling intensity of one plot per 4.95 ha (approximately, 1:5), totaling 220 plots. Data were splitted into two groups: 80% for semivariograms fits and 20% for validation. Sampling units representing four different sample intensities (1:5, 1:10, 1:15 and 1:20) were selected and the semivariograms models were fitted. The models were then used to estimate the volumes per hectare of the parcels intended for validation. It was verified that the spherical models adjusted for the different sample intensities presented satisfactory and similar performances with each other, with mean relative errors lower than 10%. The lowest value was presented in the sampling intensity of 1:5, 7.33%, and the highest in the intensity of 1:20, 8.90%. An average relative error difference of only 1.57%. Therefore, it is possible to reduce the sample intensity in pre-cut inventories for Eucalyptus stands, without great losses in accuracy, with the use of point kriging to obtain the volume of wood per hectare in non-sampled points.