Investigation of the Relationship between Gas Production and Sediment Properties in River Environments

Abstract

In riverine environments under anaerobic conditions, methane and carbon dioxide are produced as a result of biological activity, causing degradation of organic matter. Under aerobic conditions, the bacteria present degrade the organic matter, whereby the concentration of dissolved oxygen may be lowered. Thus, issues experienced in the investigation area (the Port of Hamburg) are hindered construction operations, increased greenhouse gas emissions and the echo-sounding equipment used for sonic-depth finding for ships possibly showing an erroneous depth. The purpose of this investigation was to find out how gas generation and respiration relate to the basic sediment properties and what mathematical model with the highest accuracy can predict gas generation and respiration (separately), while maintaining within a given (precision) error (1%). Gas pressure was measured at the TU Delft for an incubation period of 100 days, which was later used to calculate the gas generation (mg C/g DW) with the use of the ideal gas law. Statistical methods used to analyze the data were: Pearson’s correlation coefficient, multiple regression analysis, adjusted coefficient of determination and error analysis. The results show that both gas generation and respiration have the highest Pearson’s correlation coefficient with TOC. Furthermore, in the multiple linear regression, gas generation had the highest coefficient of determination in a regression between TOC as the primary parameter and iron content (in solids) as the secondary parameter (푅2=0.91495). For respiration, it was displayed in a regression between TOC (as the primary parameter) and copper content in the solids (as the secondary parameter) (푅2=0.881). This concludes that organic matter degradation is driven by the quantity of organic matter. The residual sum of squares showed a decrease from the linear (and non-linear) model to the multiple linear regression model. The prob>|t| value (which determines the probability of error for the multiple linear regression) was much lower than 1% for all parameters in both the gas generation and respiration model, so it can be deduced that the variables are contributing to the model in a statistically significant way. Together with the previously mentioned highest coefficient of determination, the most accurate model for both gas generation and respiration found in this investigation was the multiple linear regression model, although the model for gas generation presented little difference to that of the simpler non-linear model. The exponential nature of the optimal fit for the data suggests that there is a threshold. In areas with low organic matter content, the organic matter present is much less degradable, falling into the “slow” pool category. It is recommended to investigate other mathematical models further. There is a possibility of a more accurate model (possibly a combination of a linear and non-linear model) for both gas generation and respiration which can model the parameters even better (higher coefficient of determination while still remaining within the permitted range of error). Furthermore, it is recommended to find out why the samples listed in tables 6 and 10 deviate more than accepted from the calculated value.