The performance of two time series processing methods, the Whittaker method (WM) and Gaussian Process Model (GPM), were assessed for real time estimation of leaf area index (LAI) derived from AVHRR daily data at 0.05 degrees spatial resolution. The two methods were selected from an ensemble of methods, based on their ability to accept missing observations and to make short-term predictions. The performances of the two selected methods were evaluated as a function of the fraction of valid data and the length of gaps over a number of cases representing a range of temporal dynamics as well as distribution of missing observations. Results show that, when the length of gaps is smaller than 20 days and the fraction of valid data over the whole time series is lower than 50%, similar performances are achieved with the two methods with RMSE values lower than 0.25. For fraction of gaps higher than 50% or periods of gaps longer than 20 days GPM is more robust than the WM at the expenses of being more time consuming