| Abstract: | The management-oriented CICLAR lumped model for carbon dynamics and Greenhouse Gas (GHG) emission assessment, is presented. A metaheuristic calibration, through a Pareto based multi-objective particle swarm optimization (PSO), is used to automatically calibrate the model with data from the Capivari reservoir (southern Brazil). Two types of calibration are implemented: (1) with carbon dioxide (CO2), methane (CH4) flux, and carbon stock changes, and (2) with synthetic data based on a solution selected from (1). The calibration's performance was assessed by Nash-Sutcliffe and root means squared errors. Three synthetic scenarios are used to analyze the data distribution influence on calibration and GHG fluxes output. The results show that the spread of solutions is higher when the model is calibrated with less data (using only measured values) when compared to the ones obtained from the synthetic data series. Although there are differences between solutions calibrated with different scenarios, all of them characterized the reservoir, through the Global Warming Potential index (GWP), as a sinkhole of equivalent CO2. Moreover, the similarity among accumulated probability distribution obtained from those different scenarios, suggest that the model can be calibrated regardless of the temporal scopes of measurements. |
