I Introduction
The ever-increasing availability of Earth remote sensing data sets from the numerous satellite-based sensors, many of which are complementary to each other, presents many new possibilities to combine the data optimally through techniques like data merging or data fusion. Thus, we can derive complete global or regional maps of geophysical parameters for comparison with or as input into 3-D global- or regional-scale models. Often, information provided by any one individual sensor might be spatially incomplete or, by itself, insufficient in some way to generate daily global or regional maps. In such cases, judicious merging of colocated data obtained from identical or similar satellite-based instruments, followed by optimal interpolation to fill in the data gaps created by missing data, can create a consistent, reliable, and complete global or regional map of the geophysical parameter. In the first step, data from similar sensors can be optimally combined (with error estimates) to produce improved estimates of geophysical parameters [1]. Missing data from one sensor are replaced with available co-located data from another sensor, thus increasing the total data coverage spatially. In the second step, the remaining gaps in the data set (arising from physical and geometrical orbital constraints) are filled in with optimal spatial interpolation to yield complete global or regional maps for comparison with models. In this paper, we use the example of merging daily aerosol optical thickness (AOT) observations obtained from the MODerate resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua satellites.