USE OF MIXED MODELS TO ESTIMATE THE VOLUME OF INDIVIDUAL TREES IN FOREST TYPOLOGIES IN THE STATE OF AMAPÁ

• Autor(a) principal: Costa Monteiro, Beatriz
• Data de Publicação: 2021
• Outros Autores: Coelho de Abreu, Jadson, Figueiredo de Souza, Rafael Lucas, Costa dos Santos, Bruno, Rezende de Oliveira, Ingrid, Borges de Lima, Robson
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Biota Amazônia
• Texto Completo: https://periodicos.unifap.br/index.php/biota/article/view/5082
• topic: Efeitos aleatórios; Efeitos Fixos; Schumacher e Hall; modelagem estatística

The present aimes to evaluate the use of mixed models to estimate the volume of individual trees in different forest typologies in the state of Amapá. The study was developed in the State Forest of Amapá, using 30 sample units (conglomerates) randomly distributed. The model adjustments for the mixed model were performed using the Maximum Likelihood Method, using the software package nlme R. To select the best functional form of equation in the mixed form, the Akaike Information Criterion (AIC), Correlation Coefficient (ryy) between observed and estimated volume, square root of the mean error (RMSE%), bias and graphical analysis of the residues. The data of the variable DAP present extreme values, resulting in the high variance. The values of the probabilities (p-value) were significant for the three parameters in all models, confirming their importance in the model. Among the models with random factors, the model that best fit the volumetric estimation of individual trees was model 2, which presented lower values of AIC and BIC criteria. The likelihood ratio test was significant, with a value of 𝑝 (0.001), that is, the model with variance function was better fitted to the data. In general, the proportionality between the residuals of the equations was maintained. The mixed models have proved to be quite effective in modeling the volume of individual trees. Therefore, in the forest management, the application of this method has a very promising potential, especially in prediction and forecasting systems for forest planning purposes.Keywords: Fixed Effects; Random effects; Schumacher and Hall; Statistical Modeling.