Estimation of missing data points from remotely sensed datasets

This paper presents a method for estimation of missing data points in satellite measurements of CO₂ concentrations. The spatiotemporal character of the data allows application of dynamic multiresolution spatial models, to obtain such estimates in an efficient way. Development of the model is described in detail, starting with its static variant that is later extended to use time series for further improvement of estimation accuracy. Static estimates have been tested using both synthetic and real satellite data, and compared to a simple prediction method. The presented results show that the developed model provides modest accuracy gains over the basic estimation method. Preliminary testing of the dynamic model shows the added value of including the time-shifted data in the estimation procedure, although a more rigorous quantification of the improvement still needs to be performed.