Comparison of mathematical models for fitting particle-size distribution curves

• Autor(a) principal: da Silva, Euzebio Medrado
• Data de Publicação: 2004
• Outros Autores: Lima, Jorge Enoch Furquim Werneck, Rodrigues, Lineu Neiva, de Azevedo, Juscelino Antônio
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Pesquisa Agropecuária Brasileira (Online)
• Texto Completo: https://seer.sct.embrapa.br/index.php/pab/article/view/6782
• topic: soil texture; soil fractions; non-linear regression; growth curves; textura do solo; granulometria; regressão não linear; curva de crescimento

Particle-size distribution is fundamental for characterizing construction materials, soil mechanics, soil physics, sediment-flux in rivers, and others. The techniques used to determine the particle-size distribution of a sample are point-wise, demanding posterior interpolation to fit the complete particle-size distribution curve and to obtain values of specific diameters. The transformation of discrete points into continuous functions can be made by mathematical models. However, there are few studies to determine the best model to fit particle-size distribution curves. The objective of this work was to test and compare 14 different models with feasibility to fit the cumulative particle-size distribution curve based on four measured points. The parameter used to compare the models was the sum of the square errors between the measured and calculated values. The most recommendable models to fit the particle-size distribution curve, based on four discrete points, are Skaggs et al. 3P, Lima & Silva 3P, Weibull 3P, and Morgan et al. 3P, all using three fitting parameters.