• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Modelagem do crescimento e produ??o florestal com a estratifica??o de povoamentos empregando o m?todo k means / Growth and forest production shaping with the populations stratification using the k-means method

Rabelo, Luiz David Oliveira 21 February 2014 (has links)
Submitted by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2015-01-05T17:56:15Z No. of bitstreams: 2 luiz_david_oliveira_rabelo.pdf: 5072190 bytes, checksum: 9bb99d128c67520558b81b221c771ba9 (MD5) license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2015-01-05T17:56:44Z (GMT) No. of bitstreams: 2 luiz_david_oliveira_rabelo.pdf: 5072190 bytes, checksum: 9bb99d128c67520558b81b221c771ba9 (MD5) license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2015-01-05T17:57:03Z (GMT) No. of bitstreams: 2 luiz_david_oliveira_rabelo.pdf: 5072190 bytes, checksum: 9bb99d128c67520558b81b221c771ba9 (MD5) license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) / Made available in DSpace on 2015-01-05T17:57:05Z (GMT). No. of bitstreams: 2 luiz_david_oliveira_rabelo.pdf: 5072190 bytes, checksum: 9bb99d128c67520558b81b221c771ba9 (MD5) license_rdf: 23898 bytes, checksum: e363e809996cf46ada20da1accfcd9c7 (MD5) Previous issue date: 2014 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior (CAPES) / O trabalho teve como objetivo propor uma metodologia de estratifica??o de povoamentos florestais, testar o modelo de Clutter utilizando outras vari?veis independentes, empregar o modelo proposto por Oliveira em 2009 e propor uma alternativa para realiza??o de prognose do crescimento e produ??o de povoamentos com uma ou duas medi??es do invent?rio florestal. O estudo foi realizado com informa??es de parcelas permanentes de invent?rios florestais cont?nuos, conduzidos em povoamentos de clones de Eucalipto localizados no Estado de Minas Gerais. A defini??o dos estratos foi realizada pelo m?todo de minera??o de dados k-means, a partir dos par?metros de equa??es lineares simples de cada talh?o para as vari?veis de povoamento em fun??o da idade. O ?ndice de local para a classifica??o da capacidade produtiva foi calculado para cada talh?o e a sua determina??o foi feita utilizando o modelo de Schumacher. A classifica??o de talh?es com uma (36 meses) ou duas medi??es, (48 meses) foi realizada com o c?lculo da dist?ncia euclidiana para as vari?veis de povoamento do talh?o com rela??o ? m?dia do estrato definido pelo k-means. Foi ajustado para os dados de um mesmo conjunto de talh?es o modelo de Clutter e algumas alternativas com a substitui??o de vari?veis explicativas, bem como o modelo proposto por Oliveira. O desempenho dos modelos foi avaliado pela raiz quadrada do erro m?dio, bias %, AIC, BIC e gr?ficos com a dispers?o dos res?duos. Com o intuito de verificar se o m?todo k means foi eficiente na forma??o de estratos, procedeu se com o teste White para o ajuste dos modelos de ?rea basal e volume. O modelo de Clutter apresentou estimativas volum?tricas precisas com a estratifica??o usando o m?todo k-means. N?o verificou-se a viola??o de homocedasticidade de vari?ncias para a maioria dos estratos gerados pelo m?todo k-means. A altura total m?dia explicou com melhor desempenho a varia??o volum?trica do povoamento. A classifica??o de talh?es florestais com a dist?ncia euclidiana entre as vari?veis de povoamento apresentou resultados diferentes quanto a idade de refer?ncia. As estimativas projetadas com a classifica??o aos 48 meses obtiveram resultados mais consistentes do que as projetadas com 36 meses. / Disserta??o (Mestrado) ? Programa de P?s-Gradua??o em Ci?ncia Florestal, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2014. / ABSTRACT The work had as object to propose a forest populations stratification methodology, to test the Clutter model using other independent variables, to employ the model proposed by Oliveira in 2009 and to propose an alternative to the performance of growth prognosis and populations productions with one or two forest inventory measurements. The study was performed with permanent parcels of continuous forest inventories, conducted in Eucalyptus clones populations located in Minas Gerais. The stratums definition was performed using the method of data mining k-means, from the simple linear equations parameters of each plot for the population variables according to the age. The local index to classify the productive capability was calculated for each plot and its determination was done using the Schumacher model. The plot classification with one (36 months) or two measurements (48 months) was made with the calculus of the Euclidian distance for the plot population variables in relation to the stratum average determined by the k-mean method. The Clutter model was adjusted for the data of the same group of plots and some alternatives with the substitution of instructive variables, and also with the model proposed by Oliveira. The models development was evaluated by the square root of the average error, bias %, AIC, BIC and graphics with residues dispersion. Intending to verify if the k-means method was efficient in the stratums formation, they continued to the White test for the adjustment of the basal area and volume models. The Clutter model has presented precise volumetric estimates with the stratification using the k-means method. Homoscedasticity variances violation was not shown for the majority of the stratums generated by the k?-means method. The average total height has better explained the population volumetric variation. The forest plots classification with the Euclidian distance among the population variables has shown different results in relation to the reference age. The estimates projected with the classification at 48 months have gotten more consistent results than the ones projected at 36 months.

Page generated in 0.0431 seconds