Data about the Earth, like in many other domains, are too difficult to access. In order to perform some insight-gaining task a series of steps has to be performed which often require a spectrum of detail technology skills which are not related to the original Earth science task on hand. Special-purpose file formats with sometimes rather peculiar mechanics, juggling with horizontal, vertical, and time reference systems, and scaling up processing to large amounts of data are just a few of such common issues. One reason is that data often are provided in a more generator-centric (where generator can be a sensor or a program, such a weather forecast) than user-centric manner, which might be called “too upstream”.As is well-known, this hinders EO exploitation significantly, making such tasks impossible to conquer for non-experts and tedious for experts. For the desirable, user-friendly opposite the term Analysis-Ready Data (ARD) has been coined by the USGS Landsat team and has gone viral since.However, despite significant work, such as in CEOS, and visible progress – ultimately it is by no means clear what ARD exactly means and how it can be achieved.In this paper, we take a fresh look focusing on spatio-temporal raster data, i.e., datacubes, modeled as coverages according to the authoritative OGC and ISO standards. The Holy Grail of this study is automatic data fusion of Earth data. Based on long-term own practice (and suffering) we list shortcomings and propose ways forward, including research and standardization directions.
Read full abstract