A methodology to estimate glass eel (Anguilla anguilla L.) abundance on a daily basis containing sparse data is proposed. Our analyses are based on scientific in situ campaigns data coming from a sampling protocol which leads to spatio-temporal blanks in the fish distribution along the estuary and the time period that must be taken into account in fish abundance estimate calculated using the sparse data. Estimates at each catch locations are calculated and then extended to the whole section of the river from a Bayesian extrapolation approach which leads to a spatially explicit method. The method proposed a resulting estimator assessing a non-linear model of current speed (a crucial element for the behaviour of this passive fish) and the sampling design jointly being able to estimate the abundance of glass eel migrating during a given day in the presence of sparse data. Confidence intervals are also proposed calculated using the sensitivity of the biomass estimates to the statistical methodology and the choice of spatial extrapolation.