GLM regression applied for the analysis of quality paremeters in composting agroindustrial wastes
Keywords:
compost, generalized linear models, organic amendments, regressionAbstract
The large amount of organic waste obtained from agro-industrial production processes are very important materials for organic agriculture, as they can improve the physical, chemical and biological quality of soils. For this reason, a trial was carried out at the “La Troncal” mill, Guayas-Ecuador (2009-2010), to test three combinations of agro-industrial residues; establishing a multifactorial experimental design and examining three different combinations of filter cake, molasses and ash; two sources of microorganisms; and two types of aeration. Regression (Generalized Linear Models) was applied, through the stepwise method, of posterior elimination, to model the relationship between the dependent variable (carbon nitrogen C/N ratio) and the independent or predictive ones: Height, Organic Matter content, Conductivity, Percentage of Organic Matter and Ash in the formula, Micro-Organisms, Temperature, pH and Aeration. The final model was: E[C/N] = -10566.1 + 66.5738*Altura - 0.19824*Altura2 - 1.8069*Altura*MOCromat - 0.4597*Altura*Temp - 0.2226*Conduct + 0.0015*Conduct*F%Cenizas + 0.0039*Conduct*Temp - 18.411*F%MO + 0.3609*F%MO*Temp + 32.0059*MicroOrgC + 188.788*MOCromat + 299.196*Temp - 2.7438*Temp2 + 27.0893* pH - 90.2597*Aireac + ε.
Which turned out to be a good explanatory and predictive model to measure dependency and estimate the possible values that the C/N ratio takes based on the initial values of the independent variables. The temperature parameter is the most critical, since compost production is a dynamic process, similarly to pH and conductivity. The probability of obtaining the best compost according to the Monte Carlo method is greater when placing between 33% and 35% OM, 19% Ash and 100% of the pile is completed with Humidity; applying commercial microorganisms; using turning as aeration method; and having at the beginning of the process the conductivity in values between 2750 and 2850 µS/cm; around 55ºC of temperature and 7.5 of pH.
