Uncertainty analysis of a spatiotemporal model for submerged vegetation colonization

Abstract

This work presents an uncertainty analysis applied to the results of an ecological model. This model describes the development of submerged macrophytes colonization in a brazilian reservoir, between Sao Paulo and Parana states. To build the model we map the submerged vegetation with hydroacoustic technique to estimate submerged canopy height. Data about the light penetration into the water were also collected in some points. The dynamic model was elaborated with two variables: depth and attenuation coefficient (kt). Monte Carlo technique was used to evaluate how the existing uncertainty in the data acquisition process and measurement tools, propagated to the kriging interpolation, affects the model results. It was possible to evaluate the model output histograms, and the Root Mean Square Error (RMSE) of each simulated point in relation to the observed one. The confidence intervals were also calculated with the 5th and 95th percentiles. With this uncertainty analysis, the interval time and the points with the lowest uncertainty could be identified.