Declined peat heterotrophic respiration as consequences from zeolite amendment simulation: coupling descriptive and predictive modelling approaches

Abstract

Nowadays, halting greenhouse gasses (GHG) emission is the world's major concern to mitigate global climate change. In oil palm cultivated tropical peatland, GHG emission is primarily constituted of CO2 flux emitted from aerobic heterotrophic respiration (Rh), the natural degradation process of organic material in an oxidative environment. By coupling descriptive and predictive statistical approaches, this study attempt to gain an in-depth understanding of the effects of zeolite rates and incubation time on CO2 emission that came from aerobic Rh in peat, as well as their decomposition process. This study found that zeolite amelioration up to 30% of the peat at field capacity and starting from the first month of observation (month-1) significantly restricted peat Rh, denoted by a reduced amount of observed CO2 flux (0.021 and 0.019-0.012 mg m-2 sec-1, respectively). Both factors and several soil variables exhibited some non-linear relationships with Rh at different magnitudes and importance, showing the limitation of the traditional linear-based approach to interpreting their complex interrelationships, as well as predicting CO2 flux. This study highlights the vital role of a polynomial (GAM) and artificial intelligence (Cubist and GBM) -based pedotransfer models in improving our understanding regarding the dynamic of the peat decomposition process as affected by zeolite amendment.

Keywords

• artificial intelligence
• co2 emission
• machine learning
• multivariate analysis
• pedotransfer modelling