Page 179 - Contributed Paper Session (CPS) - Volume 3

CPS1985 Markus Z.

Figure 1: Spatial Resolution vs Revisit Time for various example satellites
(Source: Satellite Applications Catapult 2017, https://sa.catapult.org.uk/ ).

As a conclusion at this point, we can summarize that in principle NSIs are
able to produce smart business cycle statistics nearly in real time and for cross-
border regions. But for regular SBCS products, statistical offices have to do a
lot of own research and to establish cooperations with other specialized
institutions, sometimes even with private data producers.

3. Challenges for NSIs by producing trusted smart statistics
Producing the entire SBCS in statistical offices is challenging. The
challenges include the IT infrastructure, the soft- as well as at the hardware.
High resolution satellite data are really big data. For example the sentinel
satellites, from the quality point of view not good enough for SBCS, collect
images of the whole earth surface every five days. Apart from the complexity
of the data volume, the skills necessary for detecting objects in satellite
images, based on machine learning algorithms (Xu et al 2018), are not yet
available in most statistical offices. It is to discuss whether the necessary
infrastructure should be established in statistical offices. As an alternative,
semi-finished statistical products, coming from partners outside of the official
statistical system, could be integrated into official statistics because there
exists a break-even where the use of semi-finished statistical products from
private enterprises will be cheaper compared to the self-production of NSIs.
Furthermore, in some cases the self-production could be impossible because
the limitations of access to the granulated data and to the necessary skills to
transform these data into statistical products. In the case of using private data,

168 | I S I W S C 2 0 1 9